IROS 2025 Keynotes - Human Robot Interaction Session: Javier Alonso-Mora

Why It Matters

By integrating perception, planning, and whole‑body control, these methods enable robots to act safely and socially in shared spaces, unlocking large‑scale deployment across healthcare, logistics, and construction.

Key Takeaways

• Multi-agent robots need integrated perception and planning for safety.
• MPC combined with homology class paths improves global navigation decisions.
• Learning human-like policies guides robot path selection in crowds.
• Interaction-aware MPPI sampling predicts joint robot and agent behaviors.
• Geometric fabrics enable whole-body, task-adaptive control with safety guarantees.

Summary

The IROS 2025 Human‑Robot Interaction keynote by Javier Alonso‑Mora centered on the challenges and breakthroughs in multi‑agent autonomy for mobile robots. He outlined how robots must not only navigate complex, dynamic environments but also cooperate with other robots and humans while guaranteeing safety. The talk covered three research pillars: semantic‑enhanced mapping, advanced motion planning, and coordinated task allocation.

Alonso‑Mora highlighted several technical advances. By fusing geometric maps with semantic labels, robots gain richer context for planning. Model‑Predictive Control (MPC) is extended beyond a single local optimum: multiple homology‑class trajectories are generated, ranked, and the best is tracked, yielding more globally aware navigation. Human‑crowd data train a neural policy that selects the most socially compliant global path, while Interaction‑aware MPPI samples joint control inputs for the robot and surrounding agents, producing interaction‑consistent predictions. For manipulation, geometric fabrics compose differential‑equation‑based behaviors—goal reaching, collision avoidance, and whole‑body coordination—allowing fast, reactive control of mobile manipulators.

Concrete demos illustrated these ideas. A self‑driving car avoided a pedestrian using MPC, a quadrotor team carried a payload through an urban canal via decentralized MPPI, and a supermarket robot performed pick‑and‑place tasks while a human moved nearby, leveraging geometric fabrics for safety. Learning‑from‑demonstration experiments—ten tomato‑picking demos—trained a neural trajectory generator, which was then safely executed by overlaying geometric‑fabric avoidance. Decentralized vessel navigation in lakes and canals showcased scalable multi‑robot coordination without a central controller.

The implications are profound: robots that reason about human intent and jointly predict agent behavior can operate safely in public spaces, hospitals, and factories. Scalable, decentralized coordination expands deployment possibilities for swarms of drones or service robots, accelerating the transition toward truly collaborative autonomous systems in smart cities and industry.

Original Description

"Keynote Title: ""Multi-Agent Autonomy: from Interaction-Aware Navigation to Coordinated Mobile Manipulation""

Speaker Biography

Javier Alonso-Mora is a Full Professor at the Cognitive Robotics Department of the Delft University of Technology, where he leads the Autonomous Multi-Robots Lab. He received his Ph.D. degree from ETH Zurich, in partnership with Disney Research Zurich, and he was a Postdoctoral Associate at the Massachusetts Institute of Technology. His research centers on autonomous mobile robots, with a focus on navigation, motion planning, learning, and control. Key applications include mobile manipulation, autonomous vehicles, on-demand mobility, and multi-robot coordination in dynamic, human-shared environments. He co-chairs the IEEE RAS TC on Multi-Robot Systems, serves as associate editor for IEEE Transactions on Robotics, Springer Autonomous Robots and several conferences, and he was a local organizer of RSS 2024. He is the recipient of a talent scheme VENI award from the Netherlands Organisation for Scientific Research (2017), the ICRA Best Paper Award on Multi-robot Systems (2019), an ERC Starting Grant (2021) and the IEEE T-ASE Best Paper Award (2024). His work on ride-pooling has led to a commercial company, The Routing Company.

Abstract

In the pursuit of scalable, socially aware, and safety-critical autonomous systems, our recent research has focused on integrating learning, planning, and control across aerial, ground, and maritime robotic platforms. Central to this effort is the fusion of model-based and data-driven approaches, enabling robust decision-making in dynamic and uncertain environments, seamless multi-robot coordination, and the ability to learn from human demonstrations. This talk will highlight recent advances in three key areas: 1) interaction-aware navigation among other robots and humans, using sampling-based model predictive control, socially compliant behavior learning, and semantic mapping; 2) real-time task and motion planning for teams of mobile manipulators through expert demonstrations, physically grounded plans, and whole-body control; and 3) decentralized 6-DoF manipulation of cable-suspended loads by a team of drones using multi-agent reinforcement learning. These contributions advance the frontier of scalable autonomy in dynamic, multi-agent environments across diverse robotic platforms."