[Paper Review] Using Contact to Increase Robot Performance for Glovebox D&D Tasks.

Aykut Özgün Önol, Philip Long+1 more|arXiv (Cornell University)|Jul 11, 2018

Robotic Locomotion and Control20 references4 citations

TL;DR

This paper proposes a contact-implicit motion planning framework that enables a humanoid robot to maintain balance during heavy object manipulation in confined glovebox environments by dynamically generating support forces through strategic contact with the environment. Using linear complementarity constraints and nonlinear optimization, the method computes joint displacements and support forces to keep the zero-moment point within the support polygon, successfully enabling stable dual-arm manipulation in 2.5D quasi-static simulations with a position error of 0.1 m.

ABSTRACT

Glovebox decommissioning tasks usually require manipulating relatively heavy objects in a highly constrained environment. Thus, contact with the surroundings becomes inevitable. In order to allow the robot to interact with the environment in a natural way, we present a contact-implicit motion planning framework. This framework enables the system, without the specification in advance of a contact plan, to make and break contacts to maintain stability while performing a manipulation task. In this method, we use linear complementarity constraints to model rigid body contacts and find a locally optimal solution for joint displacements and magnitudes of support forces. Then, joint torques are calculated such that the support forces have the highest priority. We evaluate our framework in a 2.5D, quasi-static simulation in which a humanoid robot with planar arms manipulates a heavy object. Our results suggest that the proposed method provides the robot with the ability to balance itself by generating support forces on the environment while simultaneously performing the manipulation task.

Motivation & Objective

• To address the challenge of maintaining robot stability during heavy object manipulation in confined, contact-constrained glovebox environments.
• To eliminate the need for pre-specifying contact modes by enabling the robot to autonomously make and break contacts with the environment.
• To develop a motion planning system that prioritizes balance through support force generation while performing manipulation tasks.
• To evaluate the feasibility of using humanoid robots like Valkyrie for safe, autonomous decommissioning in nuclear facilities.

Proposed method

• Model rigid body contacts using linear complementarity constraints (LCCs) to represent contact forces and non-penetration conditions.
• Relax the equality constraint in LCCs into an inequality to improve numerical stability and allow for slack variables.
• Formulate a nonlinear optimization problem that minimizes position error, control effort, and relaxation slack, subject to grasp, ZMP, and stability constraints.
• Solve for optimal joint displacements and support force magnitudes that maintain the zero-moment point within the safe region of the support polygon.
• Implement a multi-objective torque controller that prioritizes support force generation and projects object wrench into the null space of support forces.
• Use 2.5D quasi-static simulations to evaluate the framework with a humanoid robot manipulating a heavy object on an elevated plane.

Experimental results

Research questions

• RQ1 Can a contact-implicit motion planning framework maintain robot stability during dual-arm manipulation in a confined glovebox environment without pre-specifying contact modes?
• RQ2 How effectively can the robot use environmental contacts (e.g., port walls) to shift the zero-moment point into the safe region of the support polygon?
• RQ3 What is the impact of support force distribution on joint torque requirements and system stability during object transport?
• RQ4 How does the system perform under increasing object displacement from the robot's base, particularly when the ZMP threatens to leave the support polygon?

Key findings

• The robot successfully transported a heavy object along a desired straight path with a maximum position error of 0.1 m after the initial configuration.
• The zero-moment point (ZMP) remained within the safe region of the support polygon throughout the task due to dynamically generated support forces.
• Contact with the glovebox ports was used strategically: the right arm contacted the left end of the port to shift the ZMP inward, and later the left arm contacted the right end of the left port to maintain balance.
• Support forces were significantly larger than the object wrench, making joint torques primarily dependent on support force magnitude and distribution.
• The ZMP was most centralized in step 7 due to symmetric support forces from both arms, confirming the effectiveness of balanced contact distribution.
• The magnitude of support forces and joint torques showed nearly linear dependence on the object's distance from the robot's base, indicating predictable system behavior.