Socially Adaptive Path Planning Based on Generative Adversarial Network

TL;DR

The paper tackles socially aware mobile robot navigation under diverse, uncertain human interaction by fusing Generative Adversarial Networks with RRT*-based planning. It introduces GAN-RRT*, a learning-based path planner where GAN_g outputs node costs and GAN_d enforces realistic decisions, and GAN-RTIRL, an inverse reinforcement learning framework that refines the planner from demonstration paths. The approach yields higher homotopy between planned and demonstration paths and more anthropomorphic trajectories, demonstrated through simulations and real-world user studies with improved pedestrian comfort. This work advances generalization in social navigation and provides a scalable, data-driven route synthesis for dynamic human-robot environments.

Abstract

The natural interaction between robots and pedestrians in the process of autonomous navigation is crucial for the intelligent development of mobile robots, which requires robots to fully consider social rules and guarantee the psychological comfort of pedestrians. Among the research results in the field of robotic path planning, the learning-based socially adaptive algorithms have performed well in some specific human-robot interaction environments. However, human-robot interaction scenarios are diverse and constantly changing in daily life, and the generalization of robot socially adaptive path planning remains to be further investigated. In order to address this issue, this work proposes a new socially adaptive path planning algorithm by combining the generative adversarial network (GAN) with the Optimal Rapidly-exploring Random Tree (RRT*) navigation algorithm. Firstly, a GAN model with strong generalization performance is proposed to adapt the navigation algorithm to more scenarios. Secondly, a GAN model based Optimal Rapidly-exploring Random Tree navigation algorithm (GAN-RRT*) is proposed to generate paths in human-robot interaction environments. Finally, we propose a socially adaptive path planning framework named GAN-RTIRL, which combines the GAN model with Rapidly-exploring random Trees Inverse Reinforcement Learning (RTIRL) to improve the homotopy rate between planned and demonstration paths. In the GAN-RTIRL framework, the GAN-RRT* path planner can update the GAN model from the demonstration path. In this way, the robot can generate more anthropomorphic paths in human-robot interaction environments and has stronger generalization in more complex environments. Experimental results reveal that our proposed method can effectively improve the anthropomorphic degree of robot motion planning and the homotopy rate between planned and demonstration paths.

Figures (11)

• Figure 1: A comparison of the demonstration path and two planned paths. The demonstration path is generated by the volunteer controlling the robot and the planned path are generated by RRT and A*, respectively. It is noteworthy that the demo path effectively avoid pedestrians comparing with other two paths.
• Figure 2: Description of the proposed GAN-RTIRL framework.
• Figure 3: The features of interaction between the robot and the pedestrian.
• Figure 4: The architecture of the generator. The input layer is the feature information of the map, goal and people detected by the robot. The output layer is the cost value of the current node of the robot.
• Figure 5: The architecture of the discriminator. The input layer is the feature information and cost value of the current node. The output layer is the discriminant result of the current node.

Theorems & Definitions (1)

• Definition 1