Interview with Jean Pierre Sleiman, writer of the paper “Versatile multicontact planning and management for legged loco-manipulation”

Image from paper “Versatile multicontact planning and management for legged loco-manipulation“. © American Affiliation for the Development of Science

We had the possibility to interview Jean Pierre Sleiman, writer of the paper “Versatile multicontact planning and management for legged loco-manipulation”, not too long ago printed in Science Robotics.

What’s the subject of the analysis in your paper?
The analysis subject focuses on growing a model-based planning and management structure that allows legged cellular manipulators to deal with numerous loco-manipulation issues (i.e., manipulation issues inherently involving a locomotion factor). Our examine particularly focused duties that might require a number of contact interactions to be solved, somewhat than pick-and-place functions. To make sure our method shouldn’t be restricted to simulation environments, we utilized it to resolve real-world duties with a legged system consisting of the quadrupedal platform ANYmal geared up with DynaArm, a custom-built 6-DoF robotic arm.

May you inform us concerning the implications of your analysis and why it’s an fascinating space for examine?
The analysis was pushed by the need to make such robots, specifically legged cellular manipulators, able to fixing a wide range of real-world duties, resembling traversing doorways, opening/closing dishwashers, manipulating valves in an industrial setting, and so forth. A regular method would have been to deal with every job individually and independently by dedicating a considerable quantity of engineering effort to handcraft the specified behaviors:

That is usually achieved by way of the usage of hard-coded state-machines by which the designer specifies a sequence of sub-goals (e.g., grasp the door deal with, open the door to a desired angle, maintain the door with one of many ft, transfer the arm to the opposite aspect of the door, move by way of the door whereas closing it, and many others.). Alternatively, a human knowledgeable might show the best way to remedy the duty by teleoperating the robotic, recording its movement, and having the robotic study to imitate the recorded conduct.

Nonetheless, this course of may be very gradual, tedious, and vulnerable to engineering design errors. To keep away from this burden for each new job, the analysis opted for a extra structured method within the type of a single planner that may robotically uncover the mandatory behaviors for a variety of loco-manipulation duties, with out requiring any detailed steering for any of them.

May you clarify your methodology?
The important thing perception underlying our methodology was that all the loco-manipulation duties that we aimed to resolve may be modeled as Job and Movement Planning (TAMP) issues. TAMP is a well-established framework that has been primarily used to resolve sequential manipulation issues the place the robotic already possesses a set of primitive abilities (e.g., choose object, place object, transfer to object, throw object, and many others.), however nonetheless has to correctly combine them to resolve extra advanced long-horizon duties.

This attitude enabled us to plan a single bi-level optimization formulation that may embody all our duties, and exploit domain-specific information, somewhat than task-specific information. By combining this with the well-established strengths of various planning methods (trajectory optimization, knowledgeable graph search, and sampling-based planning), we had been capable of obtain an efficient search technique that solves the optimization downside.

The principle technical novelty in our work lies within the Offline Multi-Contact Planning Module, depicted in Module B of Determine 1 within the paper. Its total setup may be summarized as follows: Ranging from a user-defined set of robotic end-effectors (e.g., entrance left foot, entrance proper foot, gripper, and many others.) and object affordances (these describe the place the robotic can work together with the thing), a discrete state that captures the mixture of all contact pairings is launched. Given a begin and aim state (e.g., the robotic ought to find yourself behind the door), the multi-contact planner then solves a single-query downside by incrementally rising a tree through a bi-level search over possible contact modes collectively with steady robot-object trajectories. The ensuing plan is enhanced with a single long-horizon trajectory optimization over the found contact sequence.

What had been your foremost findings?
We discovered that our planning framework was capable of quickly uncover advanced multi- contact plans for numerous loco-manipulation duties, regardless of having supplied it with minimal steering. For instance, for the door-traversal state of affairs, we specify the door affordances (i.e., the deal with, again floor, and entrance floor), and solely present a sparse goal by merely asking the robotic to finish up behind the door. Moreover, we discovered that the generated behaviors are bodily constant and may be reliably executed with an actual legged cellular manipulator.

What additional work are you planning on this space?
We see the introduced framework as a stepping stone towards growing a completely autonomous loco-manipulation pipeline. Nonetheless, we see some limitations that we intention to deal with in future work. These limitations are primarily related to the task-execution section, the place monitoring behaviors generated on the premise of pre-modeled environments is just viable beneath the belief of a fairly correct description, which isn’t at all times simple to outline.

Robustness to modeling mismatches may be vastly improved by complementing our planner with data-driven methods, resembling deep reinforcement studying (DRL). So one fascinating route for future work can be to information the coaching of a sturdy DRL coverage utilizing dependable knowledgeable demonstrations that may be quickly generated by our loco-manipulation planner to resolve a set of difficult duties with minimal reward-engineering.

In regards to the writer

Jean-Pierre Sleiman obtained the B.E. diploma in mechanical engineering from the American College of Beirut (AUB), Lebanon, in 2016, and the M.S. diploma in automation and management from Politecnico Di Milano, Italy, in 2018. He’s at the moment a Ph.D. candidate on the Robotic Techniques Lab (RSL), ETH Zurich, Switzerland. His present analysis pursuits embody optimization-based planning and management for legged cellular manipulation.