Research Analyzer
← Back ICRA 2026

Nonlinear Model Predictive Control for Robotic Pushing of Planar Objects with Generic Shape

Sara Federico, Marco Costanzo, Marco De Simone, Ciro Natale

PDF

AI summary

Key figure (auto-extracted from paper)
A continuous nonlinear MPC enables real-time, disturbance-robust pushing of arbitrarily shaped planar objects without hybrid contact modeling or linearization.
Nonlinear MPC Robotic pushing B-spline modeling Contact dynamics Real-time control Non-prehensile manipulation

Problem

Manipulating objects in cluttered, dynamic scenes requires precise, time-constrained pushing that maintains contact despite shape variability and occlusions, which traditional linear or hybrid controllers struggle to handle.

Approach

The method uses B-splines to model arbitrary object contours and formulates a continuous nonlinear MPC that directly optimizes pusher velocity while adapting tangential speed limits based on local curvature to prevent contact loss.

Key results

  • B-spline contour parameterization for arbitrary planar shapes
  • Continuous NMPC eliminating mixed-integer contact switching
  • Real-time solver execution on physical hardware
  • Superior tracking and disturbance rejection versus linear MPC

Why it matters

Provides a robust, model-based control framework for autonomous non-prehensile manipulation in complex, time-sensitive robotic tasks.

Abstract

Robotic manipulation of objects in cluttered dynamic scenes is challenging for a twofold reason. Object detection and localization are complex due to partial occlusions and high vari- ability in the object classes and manipulation in tight spaces is difficult due to potential collisions. The present letter focuses on the low-level control of the non-prehensile pushing action aimed at moving planar objects of generic shape along a given path with an assigned time law. Based on the continuous and nonlinear dynamics of the system, we propose a nonlinear model predictive controller (NMPC), which avoids the need for linearization and, thus, the hybrid dynamics arising from it. An extensive comparison with a state-of-the-art linear MPC demonstrates that the NMPC can successfully react to more general disturbances, outperforming the linear one. Experimental results confirm the effectiveness of the method in a task where a robot is required to grasp fruits in a container with other obstructing objects (shown in the attached video).

Index terms

Contact Modeling Dexterous Manipulation Optimization and Optimal Control

Related papers