Robotics

Towards Multi-Object Nonprehensile Transportation via Shared Teleoperation: A Framework Based on Virtual Object Model Predictive Control

A shared-control robot can transport multiple objects on a tray without dropping or tipping them.

Deep Dive

A research team from multiple institutions has published a paper detailing a novel framework for shared teleoperation, enabling a human and robot to collaboratively transport multiple loose objects on a tray. The core innovation is a division of labor: the human operator focuses on directing the tray's overall position and path, while the robot's autonomy handles the critical, high-frequency task of adjusting the tray's orientation in real-time. This addresses a major challenge in "nonprehensile" (non-grasping) manipulation, where objects can slide or tip over during rapid movement.

The team's key technical contribution is a Virtual Object (VO) Model Predictive Control (MPC) algorithm. This method simplifies complex dynamic constraints by modeling the entire collection of objects as a single virtual mass, making real-time trajectory planning feasible. An MPC-based smoothing algorithm then enforces these constraints, coordinating the user's positional commands with the necessary tilt adjustments. The system was validated in experiments where it successfully transported nine objects at accelerations up to 2.4 m/s². Compared to a baseline method, it achieved a 72.45% reduction in object sliding and completely eliminated tip-overs (0% vs. 13.9%), demonstrating robust performance in complex, dynamic scenarios.

Key Points
  • The shared-control framework splits positioning (human) and orientation (robot) tasks for stable multi-object transport.
  • Its Virtual Object MPC algorithm reduced object sliding by 72.45% and cut tip-overs to 0% in tests with nine objects.
  • The system operated at accelerations up to 2.4 m/s², proving adaptability for dynamic, real-world logistics and handling tasks.

Why It Matters

This advances robotic automation in warehouses and factories, enabling faster, more reliable transport of multiple items without complex gripping mechanisms.