Research
Synopsis
Reaching human like dexterity and agility in unstructured environments will remain a core focus area in robotics research for years to come. Humans are slower, less accurate, and precise when compared to modern collaborative robots, yet robots are far from achieving human proficiency. This can be attributed to reactiveness, geometrical awareness of environment, intrinsic embodiment conscience, which help humans cope up with changes in their surroundings. As robots get more integrated in our everyday habitats, thereby also sharing workspaces in some cases, a two fold quest, in this aspect, is necessary: (i) Embedding effective behaviors to address human-centric challenges. (ii) Defining tasks and motion policies in these fast-changing environments while still being inherently safe. To this end, the goal of my research is to develop control-based motion generation and manipulation strategies, pivoted on exploiting geometric representations like dual quaternions, for high dimensional robotic systems, and reciprocally, to use these (real-time) approaches for safe and fluid physical human-robot-interaction in dynamic and partially unknown environments.
More specifically, I seek the answers to the following research questions:
(a) How to adapt task plans effectively and be inherently reactive in dynamic scenarios even for robots with significantly large DoFs ? How to handle desired and undesired interactions with the environment/humans in this context ?
(b) How to develop methods for non-expert users to teach robots specific tasks which are complex in cluttered environments ? How to plan in clutter in the most efficient way, i.e., exploring maximum amount of information from the end user while still satisfying different robot embodiment ?
Theses
Task-specific Motion Planning using User Guidance, Imitation, and Self-Evaluation [PDF]
MS Thesis, Stony Brook University, May 2018
Enhanced Dexterity Maps (EDM): A New Map for Manipulator Capability Analysis.
Haowen Yao, Riddhiman Laha, Luis F.C. Figueredo, and Sami Haddadin
IEEE Robotics and Automation Letters (RA-L), December 2023
Predictive Multi-Agent based Planning and Landing Controller for Reactive Dual-Arm Manipulation. [PDF] [Code]
Riddhiman Laha*, Marvin Becker*, Jonathan Vorndamme*, Juraj Vrabel, Luis F.C. Figueredo, Matthias A. Müller, and Sami Haddadin
IEEE Transactions on Robotics (T-RO), December 2023
Conferences
S*: On Safe and Time Efficient Robot Motion Planning. [PDF]
Riddhiman Laha, Wenxi Wu, Ruiai Sun, Nico Mansfeld, Luis F.C. Figueredo, and Sami Haddadin
IEEE International Conference on Robotics and Automation (ICRA 2023), London, United Kingdom
Coordinated Motion Generation and Object Placement: A Reactive Planning and Landing Approach. [PDF]
Riddhiman Laha*, Jonathan Vorndamme*, Luis F.C. Figueredo, Zheng Qu, Abdalla Swikir, Christoph Jähne, and Sami Haddadin
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2021), Prague, Czech Republic
Point-to-point Path Planning based on User Guidance and Screw Linear Interpolation. [PDF]
Riddhiman Laha, Anjali Rao, Luis F. C. Figueredo, Qing Chang, Sami Haddadin, and Nilanjan Chakraborty
ASME International Design Engineering Technical Conferences and Computers and Information in Engineering Conference (IDETC 2021), Virtual, Online
Reactive Cooperative Manipulation based on Set Primitives and Circular Fields. [PDF]
Riddhiman Laha, Luis F. C. Figueredo, Juraj Vrabel, Abdalla Swikir, and Sami Haddadin
IEEE International Conference on Robotics and Automation (ICRA 2021), Xi'an, China
Refereed Workshops
Task-specific Motion Planning using User Guidance, Imitation and Self-Evaluation [PDF]
Riddhiman Laha and Nilanjan Chakraborty
RSS Workshop on Causal Imitation in Robotics, Carnegie Mellon University, June 2018