Follow-Me in Micro-Mobility with End-to-End Imitation Learning

2025-11-07

AI Summary

Autonomous following robots face significant challenges in balancing user comfort with navigation performance in dynamic indoor-outdoor environments. Traditional hand-tuned controllers struggle to generalize across diverse scenarios while ensuring smooth, safe, and responsive behavior. Method: This paper proposes an end-to-end imitation learning framework that explicitly optimizes for human-perceived comfort—quantified via subjective metrics—as the primary objective, replacing conventional controller design. A multi-neural-network architecture models expert human demonstrations to learn smooth, collision-avoidant, low-latency following policies, jointly optimizing key commercial metrics including temporal efficiency and path length. Contribution/Results: To our knowledge, this is the first work to integrate quantified subjective comfort directly into the imitation learning loss function. Evaluated on the DAAV autonomous wheelchair platform, the method achieves state-of-the-art following quality and has been successfully deployed in real-world assistive mobility scenarios, demonstrating robustness, practicality, and measurable improvements in user experience.

Technology Category

• Intelligent Robots: Learning & Optimization for ROB
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Autonomous micro-mobility platforms face challenges from the perspective of the typical deployment environment: large indoor spaces or urban areas that are potentially crowded and highly dynamic. While social navigation algorithms have progressed significantly, optimizing user comfort and overall user experience over other typical metrics in robotics (e.g., time or distance traveled) is understudied. Specifically, these metrics are critical in commercial applications. In this paper, we show how imitation learning delivers smoother and overall better controllers, versus previously used manually-tuned controllers. We demonstrate how DAAV's autonomous wheelchair achieves state-of-the-art comfort in follow-me mode, in which it follows a human operator assisting persons with reduced mobility (PRM). This paper analyzes different neural network architectures for end-to-end control and demonstrates their usability in real-world production-level deployments.

Problem

• Developing autonomous micro-mobility systems for crowded dynamic environments
• Optimizing user comfort over traditional robotic metrics like efficiency
• Creating smooth controllers through imitation learning for follow-me applications

Innovation

• Uses end-to-end imitation learning for control
• Compares neural network architectures for deployment
• Achieves comfort in follow-me wheelchair mode