Opportunities

Subject: From tactile sensing to compliant control of robotic hands for manipulation of soft objects in agricultural scenarios

Objective:

The objective of this research topic is to improve robotic capacity in the manipulation of soft objects such as fruits and vegetables in agricultural scenarios. This research aims to propose a framework for sensing and controlling robotic hands; more specifically, a framework that relies on tactile sensing for perception and compliant control for physical manipulation. From the application point of view, the focus will be on robotic capacities such as grasping, manipulating, and releasing soft objects with high dexterity and accuracy, while also ensuring safety during human-robot interaction. By improving such capacities through compliant interaction, this proposed Ph.D. thesis aims to improve productivity, reduce the arduousness of certain jobs, respond to the lack of manpower, and reduce waste in the agricultural industry. The research will also contribute to the development of robotic technologies in other domains such as manufacturing, healthcare, and assistive robotics. The automation of agricultural processes has become increasingly important due to the demand for sustainable food production and the shortage of agricultural labor. Mobile manipulation using compliant robotic hands offers a promising solution for handling delicate agricultural products, such as fruits and vegetables. However, the control and design of such systems pose significant challenges due to the need for dexterity, adaptability, and robustness. Traditional robotic hands and grippers are typically controlled for forceful power-grasping strategies. While in the context of real-world applications, the use of tactile sensing and compliance control becomes crucial to enable existing robotic hands to handle delicate objects safely and effectively. In this proposed Ph.D. thesis, we will develop strategies based on machine learning techniques to learn effective control policies that enable closing the loops from tactile sensing to impedance control of a robotic hand.

Methodology:

The research will involve a combination of theoretical analysis, simulation, and experimental validation. First, a thorough review of the state-of-the-art in robotic hand control will be conducted, with a focus on soft and fragile object manipulation. Then, a novel robotic hand control approach will be proposed based on tactile sensing and impedance control. This control framework will be developed based on a combination of model-based and data-driven approaches, with a focus on machine learning techniques such as reinforcement learning and learning from demonstration. In particular, we will tackle problems such as in-hand object localization, slippage detection, and manipulation in scenarios including picking, placing, stacking, and insertion. The proposed control approach will be evaluated through extensive experiments using real robotic hands and collaborative safe manipulators.

Requirements:

The candidate must have a Masters’s degree (or equivalent) in Engineering or Computer Sciences. Interests and experiences in Robotics, Control systems, and AI are preferable. Furthermore, the necessary qualifications are:

– Strong background in robotics and control with excellent knowledge of Math – Proficient programming skills and experience in C++ or Python

– English proficiency and communication skills (written and oral)

– Ability and motivation to work both independently and collaboratively

– Hands-on skills in ROS and Matlab are highly expected

General information:

– Supervisor: Faiz Ben Amar

– Possible co-supervision: Mahdi Khoramshahi

– Thesis collaborators: Andrea Cherubini, Philippe Fraisse (LIRMM, University of Montpellier)

– Host laboratory: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.

How to apply:

The candidate should send an email to mahdi.khoramshahi@isir.upmc.fr AND faiz.benamar@isir.upmc.fr with the following documents:

– CV

– Motivation letter

– Recommendation letters (or/and contact details of references)

– Academic transcripts for the master’s degree (or equivalent grades for the last two years).

Application deadline: 30/06/2023

Only shortlisted candidates will be contacted for online or in-person interviews.

Title : Document-level Machine translation: Translating Scientific Texts

Context:

This PhD will be fully founded by the ANR project MaTOS Machine Translation for Open Science(https://www.anr-matos.fr) which aims to develop new methods of automatically translating and evaluating scientific documents. The project focuses on translation between English and French, for which resources are readily available and translations are of a reasonable quality and coherence. The PhD will be co-supervised by Rachel Bawden (Inria) and François Yvon (CNRS).

Project Description:

Most so-called “neural” machine translation (MT) systems model the process of generating a target language document $y$ from a source language document $x$ by decomposing this process sentence by sentence, each sentence being translated independently of the neighboring sentences. The underlying probabilistic model takes thus the following form: $P(y|x; \theta)$, where $\theta$ represents the set of parameters of the model (for example a Transformer model [Vas17]). Once $\theta$ is learned, the generation of a translation is based on the search for the most probable translation realizing: $\arg \max_{y} P(y|x; \theta)$.

Such models are naive and ignore the multiple dependencies that exist between sentences within the same document. To overcome this deficiency, multiple alternative architectures have been proposed to integrate a discourse context $c$ into the model, leading to models of the form $P(y|x, c; \theta)$. Depending on the implementation, the context $c$ represents a few sentences before $x$, or the whole source document, or the beginning of the translation (the sentences before $y$). Several ways of encoding $c$ (with a dedicated encoder, or using the same encoder as for $x$) have been proposed in the literature. The most common such architectures, dedicated to document MT (DLTM), are described in [Mar21].

Two main obstacles make this extension difficult to implement: (a) the computational resources (memory and computation time) required to encode an extended context grow quadratically with the length of the context (for Transformer architectures); (b) learning the dependencies between $y$ and $c$ is made difficult by the relative scarcity of words for which the extended context $c$ is useful. Most studies in the DLMT framework address problem (a) and consider either approximations to the attention computation (see [Tay21] for a recent review) or alternative architectures to the Transformer model (e.g. [Gu21]) for encoding long sequences.

This thesis proposal addresses the problem of translating complete documents, focusing on a particular type of document: academic papers (articles, communications, research reports). These documents are relatively long, and are governed by rigid principles of organization and presentation specific to this genre of texts – division into sections and subsections – as well as by specific argumentative strategies: introduction of concepts and definitions, explicit reasoning to support precise demonstrations and conclusions, etc.

Scientific Objective:

The main objectives of the thesis are to ensure that the documents generated by machine translation (a) correctly reproduce the general structure of the input text; (b) display the same level of cohesion and coherence, especially in the choice of terms, as the source text; (c) faithfully reproduce the argumentative strategies (premises, deductions, conclusions) that are present in the source text, and (d) state the same general conclusions in the target language as in the source language. Other difficulties of the task, which may require special attention, include: the presence of many extra-lexicals (numbers, mathematical symbols, proper names) and non-textual parts (formulas, equations, tables, graphs). Finally, it should be noted that the methods considered will have to be adapted to a situation where monolingual data are abundant, but parallel data are extremely rare: this context suggests to consider the use of large pre-trained language models.

To achieve these goals, we will be interested for instance in architectures that exploit an extended discourse context (see references below) whether proposed for machine translation or for long document summarization [Koh22]. We will also consider planification methods that are used for automatic text generation [Pup22]. The main challenge of this line of work will be to find the best trade-offs between the algorithmic complexity of processing large contexts (for learning and inference) and the tangible benefit of these efforts as measured by the achievement of objectives (a-d).

A second line of work will focus on modelling the discourse strategies and communication goals associated with each part of the document: a simple way of approximating these goals is based on the internal structure of the documents, but more sophisticated approaches, using latent variable models, will also have to be considered.

Required Profile:

Candidates should have a Master 2 or equivalent (e.g. engineering school) in computer science (speciality artificial intelligence, machine learning or natural language processing).

Required skills:

The candidate should have a good level in programming (python), experience with neural networks and an interest in natural language processing. A good written and spoken level of English is required, and knowledge of French is preferred.

We are looking for highly motivated candidates with a strong background in NLP, machine learning and an interest in linguistics and language. Ideally, candidates should be able to show initiative, creativity and have a good eye for analysis of data and results.

More information: 

– Supervisor: François Yvon

– Possible co-director: Rachel Bawden (Inria Paris)

– Host laboratory: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.

Contact: 

– François YVON

– Email : yvon(at)isir.upmc.fr

– Send your application by e-mail, with [subject of the thesis] in the subject line, a CV and a letter of motivation.

Download this thesis offer

Subject: Physics-aware deep learning for modeling spatio-temporal dynamics

Context:

Physics-aware deep learning is an emerging research field aiming at investigating the potential of AI methods to advance scientific research for the modeling of complex natural phenomena. This research topic investigates how to leverage prior knowledge of first principles (physics) together with the ability of machine learning at extracting information from data. This is a fast-growing field with the potential to boost scientific progress and to change the way we develop research in a whole range of scientific domains. An area where this idea raises high hopes is the modeling of complex dynamics characterizing natural phenomena occurring in domains as diverse as climate science, earth science, biology, fluid dynamics, etc. This will be the focus of the PhD project.

Research Directions:

The direct application of state-of-the-art deep learning (DL) methods for modeling and solving physical dynamics occurring in nature is limited by the complexity of the underlying phenomena, the need for large amounts of data and their inability to learn physically consistent laws. This has motivated the recent exploration of physics-aware methods incorporating prior knowledge, by researchers from different communities (Willard et al. 2020, Thuerey et al. 2021). Although promising and rapidly developing, this research field faces several challenges. For this PhD project we will address two main challenges, namely the construction of hybrid models for integrating physics with DL and generalization issues which condition the usability of DL for physics.

– Integrating DL and physics for spatio-temporal dynamics forecasting and solving PDEs

In physics and many related fields, partial differential equations (PDEs) are the main tool for modeling and characterizing the dynamics underlying complex phenomena. Combining PDE models with ML is then a natural idea when building physics-aware DL models and it is one of the key challenges in the field. For now, this has been explored for two main directions: (i) augmenting low resolution solvers with ML in order to reach the accuracy of high-fidelity models at a reduced computational cost (Belbute-Perez et al. 2020, Kochkov et al. 2021, Um et al. 2020), and (ii) complementing incomplete physical models with ML by integrating observation data through machine learning (Yin et al. 2021a, Dona et al. 2022). The former topic is crucial for the entire field of numerical simulation while the latter allows for explorations beyond the current limits of numerical models. Simultaneously, the recent advances in neural operators (Li et al. 2021, Lu et al. 2021, Li et al. 2022, Yin et al. 2023) offer new methods for learning and modeling dynamics at different resolutions in space and time, providing the possibility of combining and learning multiple spatio-temporal scales within a unified formalism, a challenge in ML. A first direction of the PhD will then be to investigate physics-aware ML models by exploring the potential developments of hybrid models together with neural operators.

– Domain generalization for deep learning based dynamical models

Explicit physical models come with guarantees and can be used in any context (also called domain or environment) where the model is valid. These models reflect explicit causality relations between the different variables involved in the model. This is not the case for DL: statistical models learn correlations from sample observations, their validity is usually limited to the context of the training domain, and we have no guarantee that they extrapolate to new physical environments. This is a critical issue for the adoption of ML for modeling the physical world. Models of real-world dynamics should account for a wide range of contexts resulting from different forces, different initial and boundary conditions or different prior parameters conditioning the phenomenon. Ensuring generalization to these different contexts and environments is critical for real world applications. Surprisingly, only a few works have explored this challenging direction. In relation with the construction of hybrid models as described above, one will investigate this issue along two main directions. The first one exploits ideas from learning from multiple environments through task decomposition as in (Yin et al. 2021b, Kirchmeyer et al. 2022). This is a purely data-based approach. The second one, takes a dual perspective, relying on prior physical knowledge of the system equations and directly targets the problem of solving parametric PDEs (Huang 2022), exploiting ideas from meta-learning (Finn 2016).

Required Profile:

Master degree in computer science or applied mathematics, Engineering school. Background and experience in machine learning. Good technical skills in programming.

More information:

• Supervisor: Patrick Gallinari
• Collaboration within the framework of the thesis: INRIA team Ange, Paris; Inria team Epione, Sophia; Institut d’Alembert, Sorbonne University.
• Host laboratory: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.
• Start date: October/November 2023
• Note: The research topic is open and depending on the candidate profile could be oriented more on the theory or on the application side
• Keywords: deep learning, physics-aware deep learning, climate data, fluid dynamics, earth science

• Institution : Sorbonne University SIM (Sciences, Engineering, Medicine)
• Doctoral school : Mechanical Sciences, Acoustics, Electronics and Robotics of Paris
• Specialty: Human Machine Interaction
• Team: Multi-Scale Interactions
• Research unit: Institute of Intelligent Systems and Robotics
• Thesis supervisor : Gilles BAILLY
• Co-supervisor : David GUEORGUIEV
• Start of the thesis: October 1, 2023
• Application deadline (at 23:59): May 15, 2023

Link to the announcement and how to apply

Project description

This thesis aims to develop a new scientific instrument for applications in experimental biology, in particular for the handling, characterisation and analysis of objects such as isolated cells, neurons, or intracellular organs. Using the principle of optical tweezers, laser beams arecontrolled to act directly on samples, or to actuate remotely controlled remote-controlled microrobots in a microfluidic environment. These microrobots will be able to integrate  analysis and bio-active sensors, thus providing rapid feedback to the operator. This is a new technology capable of supporting and and accelerate several studies in biology. Collaborations have been started with teams from the Institut Curie and Pasteur on cancer diagnosis and treatment, and exploration of intracellular mechanisms.

Profile and skills required

General engineering; Robotics; Applied Physics; Scientific Instruments. Prior knowledge on experimental biology or laser optics will be a plus.

General information :

• Institution : Sorbonne University SIM (Sciences, Engineering, Medicine)
• Doctoral school : Mechanical Sciences, Acoustics, Electronics and Robotics of Paris
• Specialty: Robotics
• Team: Multi-Scale Interactions
• Research unit: Institute of Intelligent Systems and Robotics
• Thesis supervisor : Sinan HALIYO
• Co-supervisor : Stéphane REGNIER
• Start of the thesis: October 1, 2023
• Application deadline (at 23:59): May 22, 2023

Link to the announcement and how to apply

Thesis subject: Efficient interaction through information maximization

Thesis subject: Designing interaction for remote collaboration in the operating room

Context

To perform surgery, surgeons typically work in collaboration with a team of experts, for example a more senior surgeon advising on a surgical technique, a radiologist interpreting an x-ray or ct- scan, or an anatomic-pathologist interpreting microscopic images of tissue to diagnose cancer. These experts are typically not present in the Operating Room (OR), and therefore the surgeon conducts remote consultations through phone calls. Our hypothesis is that existing tools for remote collaboration are not effective nor easy to use when working in the OR because they lack effective interaction mechanisms. The challenges of interacting with a collaborative system are that the primary task, performing surgery, is cognitively demanding, but also physically demanding, as both hands of the surgeon are busy handling instruments (Avellino et al., 2021). Moreover, the hands of surgeons are sterile, and using classic means of interaction (mouse or keyboard) requires breaking sterility and re-sterilizing before going back to the patient. Previous works in Human–Computer Interaction (HCI) and Computer Supported Collaborative Work (CSCW) have studied interaction techniques in the OR, for example controlling imaging systems through voice and gesture (Feng et al., 2021; Mentis et al., 2015) or a combination of gaze and feet (Hatscher et al., 2017), and a robotic endoscope using a multimodal technique that combines several input mechanisms (Avellino et al., 2020). Nonetheless, these techniques (1) have not been studied in the context of remote collaboration, but rather individual work in the OR, and (2) they have not been adopted in real surgical work thus far, pointing to a partial support of the wide variety of needs during surgery. So far, research has been able to conduct studies where the remote expert interacts, such as studying the benefits of remote pointing (Mentis et al., 2020; Semsar et al., 2019, 2020). This project will open the door for a large body of work that can study the benefits and challenges of remote collaboration where both parties can interact with systems.

Job title: Post-doc on haptic and multisensory human-computer interactions

Context:

This post-doc is part of the NeuroHCI ANR project. The overall goal of NeuroHCI is to improve human decision making in the physical and digital worlds in interac,ve contexts. There are various scenarios in which a human makes a decision with an interac,ve system. The decision might be about a complex real-world choice assisted by a computer (e.g. medical treatment), the choice of a method to achieve a digital task (e.g. editng a photo with the preferred tool), or the way we decide the best way to perform a haptic interaction.

Missions:

The envisioned scientific approach will rely on optimizing the haptic feedback delivered to the user with relation to vision and hearing by leveraging computational models of multisensory integration. Thus, the scientific activities of the project will revolve around questions including but not limited to:

– How to ensure that the inconsistencies between what the user sees and what the user feels does not break the illusion and how to mitigate their effects on user experience?

– How visuo-haptic inconsistencies influence users’ strategies (e.g. which objects they will decide to interact with) and high-level decision making?

Required profile:

The ideal candidate must have a PhD degree and a strong background in human-computer interaction and/or cognitive science.

Required skills:

– Experience in haptics is a strong plus ;

– Strong skills in Python, Matlab or equivalent ;

– Good knowledge of experimental design, psychophysics and statistics ;

– Excellent publication record ;

– Willingness to work in a multi-disciplinary team ;

– Good communication skills.

General Information:

• Contract start date: at the latest 06/01/2024
• Contract duration: 24 months
• Working time: 100%
• Desired experience: beginner at 4 years old
• Level of studies required: PhD
• Host laboratory: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.

How to apply:

– David Gueorguiev ; david.gueorguiev(at)sorbonne-universite.fr

– Send your application by email, with [name of offer] in the subject line, a CV and a cover letter.

Download this offer

Job title: Engineer position (24 months) in explainable AI for onco-immunology at ISIR / Institut Curie

Context: The XOMX collaborative project between ISIR and Institut Curie

Missions: Development and application of explainable AI approaches for onco-immunology.

The project is in the context of a collaboration between ISIR and the Waterfall lab (https://institut-curie.org/team/waterfall/) of Institut Curie in Paris (https://www.isir.upmc.fr/) and will entail active interactions with both teams.

The main focus will be to contribute to the development of the XOMX tool suite (https://github.com/perrin-isir/xomx) for interpretable AI approaches in analyzing highthroughput molecular profiling datasets.  Specific applications include cancer diagnosis from DNA/RNA profiling, single cell sequencing analysis, and immunopeptidomics.

The successful candidate will work in a strong collaborative environment, also linked with wet-lab biologists and clinicians.

Required profile:

We primarily expect candidates with strong computer science/quantitative science backgrounds and a motivation to learn onco-immunology.

Required skills:

– Experience in python is essential and additional languages (eg R) and pipeline managers (nextflow, snakemake, kedro) are welcome.

– Experience with biological datasets (especially highthroughput sequencing data) is a plus.

– Skill to work on large computing clusters.

– Self-motivated and able to collaborate with biologists and clinicians.

– Presentation and communication of results.

General Information:

• Contract start date: as soon as possible
• Contract duration: 24 months
• Working time: 100%
• Desired experience: beginner at 10 years old
• Level of studies required: Master 2
• Host laboratory: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.

How to apply:

– Nicolas Perrin-Gilbert; perrin(at)isir.upmc.fr

– Send your application by email, with [name of offer] in the subject line, a CV and a cover letter.

– Deadline for submitting the application: 15/10/2023

Download this offer

Post-doc position “Manipulation in robotics with model-free model-based hybrid approaches”

Research activity:

The goal of the PostDoc is to advance the state-of-the-art in robotic manipulation regarding industrial and agricultural applications in the context of EU projects PILLAR and EuRobin. The main objective will be to develop control strategies to accomplish manipulation tasks. For this purpose, we will explore hybrid approaches benefiting from both model-based (such as optimal control, model predictive control, and model-based RL) and model-free ones (such as learning from demonstration, and model-free RL). The majority of the research activity for this position will be dedicated to formulating, developing, programming, validating, and finally integrating such control strategies into the dedicated robotic platforms; i.e., integrating with perception, planning, natural language, and cognitive modules that are being developed in the EU projects. The robotic platform used for developing algorithms consists of two 7-axis Franka Emika collaborative arms. The manipulation capacities targetted in these projects are diverse: pulling/pushing sliders and drawers, opening/closing doors, cleaning and tidying up working stations, handing-over objects, using tools, probing, manipulating soft objects such as cables, etc.

The research activities will be supervised by Prof. Stephan Doncieux and Prof. Mahdi Khoramshahi in collaboration with the other researchers at ISIR involved in PILLAR-robots and EuRobin projects.

The position:

This is a one-year full-time PostDoc position. A second-year contract will be granted upon completion of the first year and the satisfaction of both parties. The position will be paid according to the French salary regulations for postdoctoral scholars considering the level of experience of the candidate.

The required Skills:

The applicants should ideally have:

1) a Ph.D. in robotics and Control Systems,

2) good experience with programming (C++, Python under ROS1 and ROS2), and experience with robotic simulation environments (e.g., Gazebo and Bullet) will also be appreciated,

3) a clear publication record in the major robotics conferences/journals (e.g., ICRA, IROS, RSS, RAL, TRO, IJRR),

4) Strong interest in scientific research: both theoretical (e.g., physical human-robot interaction, intention recognition, manipulation, and grasping) and experimental (design and implementation of experiments with integrated systems and robots),

5) Ability to collaborate with high autonomy and self-responsibility

6) Availability to travel to project meetings with partners.

The PILLAR-robots project:

The EU-funded PILLAR-Robots project is developing a new generation of robots that can build on the experience acquired during the robots’ lifetime to fulfill the wishes of their human designers/users in real-life applications. Researchers will operationalize the concept of “purpose,” drawn from the cognitive sciences, to increase robot autonomy and domain independence during autonomous learning. The goal is to provide the robots with the knowledge and skills needed to operate under targeted applications. The project will use purposeful intrinsically motivated cognitive architecture in agri-food, edutainment, and unstructured industrial/retail field demonstrations.

https://cordis.europa.eu/project/id/101070381

euROBIN: A European network of excellence in robotics:

The euROBIN (European ROBotics and AI Network) project is an initiative funded by the European Union to create a network of excellence in robotics and artificial intelligence (AI). This network brings together leading researchers, institutions, and industrial partners in the field of robotics and AI, to develop innovative European technologies and solutions. The vision of euROBIN is to create a European ecosystem of robots capable of sharing their data and knowledge, exploiting their diversity to jointly learn to perform an infinite variety of tasks in human environments. The euROBIN project aims to make significant progress in four key scientific areas: Interaction with the environment, Transfer of learned knowledge, Transferable knowledge representation, and Human-centred knowledge transfer. The euROBIN project will demonstrate the relevance of its scientific results in four promising areas of application: personal robots, industrial robotics, robotics for the circular economy, and robots for quality of life and well-being.

The euROBIN network includes 31 partners from 14 countries, with leading research institutions and industrial partners in the field of robotics and AI.

https://www.eurobin-project.eu/

General Information:

• Contract start date: from September 2023
• Contract duration: 12 months
• Working time: 100%
• Desired experience: 1 to 10 years
• Level of studies required: PhD
• Salary: based on salary scale
• Host laboratory: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.

How to apply:

Interested applicants can contact Prof. Mahdi Khoramshahi [mahdi.khoramshahi@isir.upmc.fr] AND Prof. Stéphane Doncieux [stephane.doncieux@sorbonne-universite.fr], with a subject including “[POSTDOC CANDIDATE]”, providing their CV and a cover letter briefly describing their background and their career plans. The position remains open until a satisfactory candidate is found.

Download the postdoc position

Position title: Human-Aware situation assessment for joint action

The PIRoS (Perception, Interaction and Robotique Sociales) team of the Institute for Intelligent Systems and Robotics (ISIR) at Sorbonne University (Paris) is looking for a for a highly motivated and ambitious postdoctoral researcher to conduct research on human-robot interaction & machine learning.

Description

Communication is a key factor to achieve successful coordination during a joint huma-robot action. Humans and robots communicate and coordinate during the execution of the joint action using multimodal cues such as speech, gaze and gestures.  By doing so humans build a mental model of the robot. Mental models enable humans to infer a robot’s intention, anticipate actions, establish a common ground and share goals. However, endowing robots with similar models is challenging.

This post-doc position will be focused on the development of new computational models of human-robot communication. These human-aware models will be built by continuously observing human activities and environment and aim to infer human mental states. Human-Centered Machine Learning techniques will be developed to explicitly take into account human specificities in the prediction of multiple mental states such as beliefs, intentions, preferences, competence and rationality. Following a Human-Centered approach, the post-doc position will also consider ethical issues in both modeling (e.g. biases) and experimental (e.g. with human participants) parts of the research work.

Human-Aware situation assessment systems will be evaluated in collaborative tasks such as human-robot handovers using both quantitative (e.g., task efficiency) and qualitative metrics (fluency, trust). The candidate will have the opportunity to conduct experiments with various robots (Franka Emika, Pepper, Mobile Manipulators) as well as ISIR’s robots partners.

She/He will work in collaboration with PhD students, post-docs and public/private partners of ISIR. In particular, the position is part of the euRobin network, which aims to advance AI tools, software, architectures, and hardware components in a reproducible approach (European Network of Excellence Centres in Robotics (RIA)). This position is for 18 months contract, but there is a possibility to be extended depending on the performance and circumstances.

Requirements 

The ideal candidate must have a PhD degree and a strong background in machine learning, robotics or cognitive science/neuroscience.
The successful candidate should have:

• Experience in robotics
• Good knowledge of Machine Learning Techniques
• Good knowledge of experimental design and statistics
• Excellent publication record
• Strong skills in Python
• Willing to work in multi-disciplinary and international teams
• Good communication skills

Application 

Interested candidates should submit the following by email in a single PDF file to: mohamed.chetouani[@]sorbonne-universite.fr with the subject: Application Post-Doc HRI-ML

1. Curriculum vitae with 2 references (recommendation letters are also welcome)
2. One-page summary of research background and interests
3. At least three papers (either published, accepted for publication, or pre-prints) demonstrating expertise in one or more of the areas mentioned above
4. Doctoral dissertation abstract and the expected date of graduation (for those who are currently pursuing a Ph.D)

Application’s deadline: May 15, 2023.

General Information: 

• Contract start date: from July 2023
• Duration of the contract : 18 months
• Working hours: 100% of the time
• Desired experience: 1 to 10 years
• Desired level of studies: PhD thesis
• Host laboratory: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.

Contact person:

• Mohamed Chetouani
• Tel:+33144276308
• Email : mohamed.chetouani(at)sorbonne-universite.fr

Download this offer

Position title: Learning multimodal behavior representations for personalized human-machine interaction

The PIRoS (Perception, Interaction and Robotique Sociales) team of the Institute for Intelligent Systems and Robotics (ISIR) at Sorbonne University (Paris) is looking for a for a highly motivated and ambitious postdoctoral researcher to conduct research on human-machine interaction & machine learning.

Description

Personalized Human-Machine Interaction systems provide experiences that are tailored to the human partner’s individual needs and preferences. For this purpose, they require user models that are usually inferred from a user profile and/or from the observation of human’s actions. The ability to adapt to changing contexts or individuals is important and poses numerous challenges concerning multimodal data collection and interpretation, privacy, and transparency. There is a need to develop new human behavior representations able to reflect heterogeneity between users, while preserving privacy.

This post-doc will be focused on the development of human-centered machine learning techniques for personalized adaptation. These techniques will result in the computation of human behavior representations from multimodal data using pragmatic reasoning in order to improve interpretation of context-dependent components of human-behaviors. Pragmatic reasoning will equip human-machine interaction systems (robots, serious games) with a greater degree of human partner awareness enabling them to account for latent intent or state. Following a Human-Centered approach, the post-doc position will also consider ethical issues in both modeling (e.g. biases, privacy) and experimental (e.g. with vulnerable participants) parts of the research work.

To evaluate the effect of computational models on the personalization of human-interaction systems, experiments will be conducted with robots/serious games with different profiles (children, adults, seniors). The candidate will have the opportunity to take advantage of experimental settings of the team, including a Neuro-Development Living & Learning Lab s (LiLLab).

This position is for 18 months contract, but there is a possibility to be extended depending on the performance and circumstances.

Post-doc : Learning in robotics, with application to grasping

Context:

During the FET Proactive DREAM project (http://dream.isir.upmc.fr/) has been defined an approach for adaptive robotics based on open-ended learning. The main goal is to allow a robot to learn without requiring a careful preparation by an expert. This approach raises many challenges, notably learning with sparse reward, representation learning (for states and actions), model learning and exploitation, transfer learning, meta-learning and generalization. These topics are considered in simulation, but also on real robotics setup, notably in the context of grasping.

Missions:

This position aims at contributing to these topics in the context of several European projects, in particular SoftManBot, Corsmal, INDEX and Learn2Grasp. Calling upon previous works in the research team, the proposed approaches need to be easy to adapt to different robotic platforms and will thus be applied to different robots (Panda arm from Franka-Emika, Baxter, PR2 or TIAGO, for instance).

Required profile:

Candidates for the position must have a PhD degree in machine learning or related field in which robotics applications (either simulated or real) have been considered.

Required skills:

An excellent background is expected in machine learning as well as an experience in robotics. Excellent programming skills in Python are expected.

General Information: 

• Position Type: Post-doctoral researcher
• Contract duration: 24 months
• Level of education required: PhD
• Remuneration : Remuneration according to experience
• Location: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.

Contact person: 

• Stephane Doncieux
• stephane.doncieux(at)sorbonne-universite.fr
• Send your application by email, with a CV and a cover letter.

Download this offer

Doctoral, Postdoctoral or research engineer positions for the HCI Sorbonne group (Human Computer Interaction)

Context:

We have multiple postdoctoral and engineering positions in the HCI group (https://hci.isir.upmc.fr) at Sorbonne Université, Paris, France.

Missions:

We are searching for curious minds who want to conduct cutting-edge research at the intersection of HCI with VR, Haptics, Robotics or AI. Possible topics/areas of research are:

• Novel Interaction techniques in VR,
• VR and haptics for gaming or training,
• Computational models for learning, decision making and human performance,
• AI-based recommendation systems.

Some of our previous work in these areas:

• CoVR (UIST21): https://dl.acm.org/doi/10.1145/3379337.3415891
• AI-based Recommendation systems (CSCW 21): https://dl.acm.org/doi/abs/10.1145/3476068?sid=SCITRUS
• Adapting UIs with Reinforcement Learning (CHI 21): https://dl.acm.org/doi/abs/10.1145/3411764.3445497
• Mixed Control of Robotic systems (CHI 20): https://dl.acm.org/doi/10.1145/3313831.3376795

Required profile:

For the postdoctoral position, a Phd degree in Computer science, HCI or other field related to our research areas is required.

Required skills:

• strong programming and analytical skills,
• strong background in at least one of the following areas (HCI, VR, Haptics, Robotics, AI).

More information : 

• Type of position: Postdoctoral or Research Engineer position
• Start date: as soon as possible
• Duration: 1 to 2 years
• Level of study required: Master 2 (for engineer), PhD (for post-doc)
• Location: ISIR (Institut des Systèmes Intelligents et de Robotique), Campus Pierre et Marie Curie, 4 place Jussieu, 75005 Paris.

Contacts: 

• Gilles Bailly et Sinan Haliyo
• Email : gilles.bailly(at)sorbonne-universite.fr ; sinan.haliyo(at)sorbonne-universite.fr
• Application: Send your application by email, with a CV and a cover letter.
• Application deadline: None

Download the offer