Learning to Throw Objects Safely in Multi-Obstacle Environments

2026-07-07

AI Summary

This work addresses the challenge of enabling a robot to safely and reliably throw objects into a target bin within cluttered environments containing stochastic obstacles. The authors propose a Potential Field-based Representation (PFR) that encodes both target attraction and obstacle repulsion into a fixed spatial grid. Combined with kinesthetic demonstration for policy initialization and off-policy reinforcement learning algorithms (SAC, DDPG, TD3), this approach substantially enhances generalization to unseen obstacle configurations. The method facilitates efficient sim-to-real transfer and demonstrates remarkable robustness and practicality, achieving up to 90% success rates in real-robot experiments with previously unencountered throwable objects. These results underscore the framework’s scalability, reliability, and applicability to real-world robotic manipulation tasks.

Abstract

Robotic throwing enables fast and efficient object placement beyond the robot's immediate workspace, but reliable throwing in cluttered environments remains underexplored. Existing approaches, such as TossingBot, learn throwing strategies from visual input but assume obstacle-free settings. In this paper, we address the problem of throwing objects into a target basket while avoiding obstacles placed randomly in the scene. We introduce a potential field state representation that compactly encodes both basket attraction and obstacle repulsion on a fixed-size grid, enabling reinforcement learning (RL) policies to generalize across arbitrary numbers and configurations of obstacles. The policy is initialized from kinesthetic demonstrations and optimized in simulation using three state-of-the-art RL algorithms (SAC, DDPG, TD3). Among these, SAC achieves the most consistent performance across scenarios. We compare the potential field representation against explicit state encodings and demonstrate that it achieves higher success rates and better scalability to unseen obstacle configurations. Real-robot experiments with unseen throwable objects confirm robust sim-to-real transfer, achieving up to $90\%$ success in cluttered scenes. These results demonstrate that PFR provides a practical and robust representation for safe and efficient robotic throwing in unstructured environments. A video showcasing our experiments is available at: https://youtu.be/ZZnJf8ua2dE

Problem

• robotic throwing
• obstacle avoidance
• cluttered environments
• safe manipulation
• target placement

Innovation

• Potential Field Representation
• Robotic Throwing
• Reinforcement Learning
• Obstacle Avoidance
• Sim-to-Real Transfer