Join us!

This exciting project focuses on developing intelligent robotic systems capable of performing complex, forceful tasks with bimanual coordination. Applications include household tasks and advanced manufacturing — For instance: Opening child-safe medicine containers, cutting, drilling, and manipulating complex objects while using the other hand for stability and adaptability.  

​

Key Highlights:
(Exploring a few of those directions according to your preference — Not all!)

• Mastering bimanual Manip.: Explore sequential planning, tool usage, and motion-task strategies;

• Robust Execution: Ensure stability and safety during real-time operations;

• Whole-body and Beyond-Body: Explore body-contacts and environment points and fixtures to stabilize objects;

• Real-Time Motion: Implement real-time controllers and planners under constraints;

• Fundamentals of Whole-Body Learning: Improve task and embodiment representation/understanding;

• Learning by Demo/Imitation Learning: Learning from demonstrations at the task or motion level. 

You must Learn: Bimanual kinematics, geometric methods, geometric and force constraint definition, basic control and motion planning.​

​​

​

Qualifications:

BSc/MSc in robotics, AI, or related fields         |      Strong programming skills (C++, Python, ML frameworks)

Teamwork and communication abilities           |      Strong passion for research, and curious personality

Desired: Experience in robotic manipulation, motion planning, and control, simulation environments, and/or use of real-robots.

For more information, check the following research papers:    

Switching strategy for flexible task execution using the cooperative dual task-space framework (ICRA, 13)

Manipulation planning under changing external forces (Autonomous Robots, 2020)

Predictive Multi-Agent based Planning and Landing Controller for Reactive Dual-Arm Manipulation (Transaction on Robotics, 2024)

This project aims to rethink and redefine how robots (and us) interact with and manipulate objects in contact-rich scenarios. This is an ambitious problem

involving geometric/Riemannian perspective linear algebra and the SOTA of ML for identification and classification. Applications are literally everywhere in

the real-world, exploring the physical dynamics of contact. For instance, simple activities such as handling multiple door handles and tilt-and-turn windows 

demand exploring the system's different articulations with limited force and sequential connection of constraints (one can only open the window once it tilts

enough). Similarly, assembly and disassembly tasks in industry often require tactile exploration followed by sequential manipulation.   

​

Key Highlights:
(Exploring a few of those directions according to your preference — Not all!)

• Task Redefinition: Frameworks for contact-rich manip. using geom. methods | Riem. manif. | intellig. feature space connecting forces/constraints;

• Geometric-Oriented Data-Analysis: for different tasks and task classification and identification with set-based constraints;

• Learning in Virtual Environment: Exploring VLMs for cues and physics engines for contacts;

• Learning in the Physical Domain: Enable task learning under set-based constraints and failure handling;

• Safety certified manipulation: in terms of constraints satisfaction (set-based methods) and torque-based methods;

• Task Exploration: Under hard safety-constraints, enlarge the soft constraints for improving the task - Touch and find;

​

What you gotta learn?!: Physical kinematics and dynamics, geometric methods, Riemannian manifolds, geometric and force constraint definition, data-analysis, and machine learning.
 

​

Qualifications:

BSc/MSc in robotics, AI, CS, Eng, or related fields      |      Strong programming skills (e.g., C++, Python, ML, Matlab)

Teamwork and communication abilities                      |      Strong passion for research, and curious personality

Desired:  Experience in robotic manipulation, motion planning, and control, simulation environments, and/or usage of real-robots. 

For more information, check the following research papers:    

• CITR: A Coordinate-Invariant Task Representation for Robotic Manipulation (ICRA, 2024)

• Integrated Bi-Manual Motion Generation and Control shaped for Probabilistic Movement Primitives (Humanoids, 20222, best paper finalist)

• Manipulation planning under changing external forces (Autonomous Robots, 2020)

• Shared Autonomy Control for Slosh-Free Teleoperation (IROS, 2023)