En poursuivant votre navigation sur ce site, vous acceptez le dépôt de cookies dans votre navigateur. (En savoir plus)

PhD on the development of an hybrid symbolic/geometric approach for planning collaborative manipulation tasks (M/F)

This offer is available in the following languages:
Français - Anglais

Date Limite Candidature : lundi 9 août 2021

Assurez-vous que votre profil candidat soit correctement renseigné avant de postuler. Les informations de votre profil complètent celles associées à chaque candidature. Afin d’augmenter votre visibilité sur notre Portail Emploi et ainsi permettre aux recruteurs de consulter votre profil candidat, vous avez la possibilité de déposer votre CV dans notre CVThèque en un clic !

Faites connaître cette offre !

General information

Reference : UPR8001-THISIM0-002
Workplace : TOULOUSE
Date of publication : Monday, July 19, 2021
Scientific Responsible name : Thierry Siméon
Type of Contract : PhD Student contract / Thesis offer
Contract Period : 36 months
Start date of the thesis : 1 October 2021
Proportion of work : Full time
Remuneration : 2 135,00 € gross monthly

Description of the thesis topic

Motion planning algorithms is an important and very active field of research in Robotics that has been rapidly progressing in recent years, notably thanks to the development of probabilistic exploration methods introduced at Stanford, to which the LAAS has contributed (e.g. [1-5] ) and which today allow a large class of combinatorially complex problems to be dealt with. Object manipulation task scheduling is a more complicated problem instance that consists not only of scheduling the collisionless movements of the robot but also the sequencing and interdependencies of the various elementary actions (e.g. picking, placing and transferring an object) that make it possible to perform the task. The work of LAAS has led to original and complementary planning techniques to solve these problems at both the geometrical [6] and symbolic [7] levels.

The thesis deals with the combination of these geometrical/symbolic planning techniques in a hybrid approach under the theme "Combined task and Motion Planning" (CTAMP) which is very popular today in the community, and which aims to better cope with the high complexity of these manipulation task planning problems on which the LAAS has a recognized expertise, both in the geometrical (motion) and symbolic (task) components. Indeed, these problems require the exploration of very high dimensional spaces that cannot be efficiently handled only at the geometric level, and moreover the solution cannot be limited to a set of partially ordered high level actions but also requires a complete instantiation of these actions that indicates the different locations of robots and objects as well as robot movements and their synchronization.

The contributions available today in the literature do not provide satisfactory answers in a reasonable time when confronted with situations as complex as those we propose to study. A promising avenue will certainly be the study of the coupling of learning methods with motion and/or task planning algorithms for a more efficient search for solutions.

We will also be interested in refined models of the interaction constraints (robot-robot or human-robot), in particular through criteria to be optimized which take into account not only the configuration of the human-robot system but also the trajectory and the dynamics of the movement, with more sophisticated cost functions, to be developed or resulting from a training of the interaction constraints on the task, and which will be able in addition to require the kino-dynamic extension of the algorithms, in order to take into account the dynamics of the movement and the interactions.

[1] Path deformation roadmaps : compact graphs with useful cycles for motion planning. L. Jaillet, T. Siméon. The International Journal of Robotics Research (2008) http://www.laas.fr/~nic/Papers/08ijrr.pdf
[2] Disassembly path planning for complex articulated objects. J. Cortés, L. Jaillet, T Siméon. IEEE Transactions on Robotics , (2008) http://www.laas.fr/~nic/Papers/08itro.pdf
[3] Sampling-based path planning on configuration-space cost maps L. Jaillet, J. Cortés, T. Siméon. IEEE Transactions on Robotics , (2010) http://www.laas.fr/~nic/Papers/10TRO.pdf
[4] Parallelizing RRT on large-scale distributed-memory architectures D. Devaurs, T. Siméon, J. Cortés. IEEE Transactions on Robotics (2013) http://hal.archives-ouvertes.fr/hal-00861579/document
[5] Optimal path planning in complex cost spaces with sampling-based algorithms. D. Devaurs, T. Siméon, J. Cortés. IEEE Transaction on Automation Science & Engineering (2016) https://hal.archives-ouvertes.fr/hal-01231482/document
[6] Manipulation planning with probabilistic roadmaps. T. Siméon, J Cortés, JP. Laumond, A Sahbani. The International Journal of Robotics Research, (2004) http://www.laas.fr/~nic/Papers/03ijrr.pdf
[7] A hybrid approach to intricate motion, manipulation and task planning.
R. Alami, F. Gravot, S. Cambon. The International Journal of Robotics Research, (2009) https://hal.laas.fr/hal-01976081/file/2009-ijrr-cambon-report.pdf

Work Context

This work will be carried out within the "Robotics and InteractionS" (RIS) group of the LAAS (https://www.laas.fr), which is developing an ambitious research theme focusing on autonomous machines integrating perception, reasoning, learning, action and reaction capacities.
The experimental supports will be the robots available at the LAAS and the experimental means of the ADREAM room.

We talk about it on Twitter!