Halvtidsseminarium - Xuezhi Niu: "Enhancing Cooperation in Heterogeneous Multi-Robot Systems: Symbiosis-Inspired Interaction of Cooperation"

Välkommen till ett halvtidsseminarium presenterat av Xuezhi Niu. Eventet är öppet för alla och hålls på engelska.

Ordförande: Per Mattsson (Avdelningen för systemteknik)

Abstract:

Heterogeneous multi-robot systems bring robots with different capabilities, resources, and roles into shared tasks. These differences can be complementary, but they also make coordination state- and task-dependent: the value of one robot’s action often depends on the state and behaviour of another. Symbiosis-inspired interaction modelling describes these directed effects and introduces them into decentralised robot decision-making.

In this half-time seminar, we will first present battery-state sharing in a two-AMR warehouse, where robots use a partner’s energy state to coordinate task execution and recharging [1]. Then we will focus on mutual-benefit reward shaping across tasks with different levels of coupling and complexity [2]. Next, we will present event-level reward shaping for tasks in which throughput depends on complementary robot roles [3]. Finally, we will introduce an online pairwise preference graph that adapts the active interaction structure without requiring full connectivity [4].

Reference: [1] X. Niu, N. C. Barajas and D. G. Broo, "Enabling Symbiosis in Multi-Robot Systems Through Multi-Agent Reinforcement Learning," 2025 IEEE 8th International Conference on Industrial Cyber-Physical Systems (ICPS), Emden, Germany, 2025, pp. 1-7, doi: 10.1109/ICPS65515.2025.11087893.

[2] X. Niu and D. G. Broo, "Investigating Symbiosis in Robotic Ecosystems: A Case Study for Multi-Robot Reinforcement Learning Reward Shaping," 2025 9th International Conference on Robotics and Automation Sciences (ICRAS), Osaka, Japan, 2025, pp. 112-117, doi: 10.1109/ICRAS65818.2025.11108729.

[3] X. Niu and D. Gürdür Broo, “PRISM: Policy-Shaping via Reward Decomposition for Inter-Agent Symbiosis in Multi-Robot Cooperation,” Manuscript under review, 2026.

[4] X. Niu and D. Gürdür Broo, “MORPH: Self-Organising Multi-Robot Task Allocation via Neuroplasticity-Inspired Adaptive Topology,” Accepted for presentation at the 23rd European Conference on Multi-Agent Systems (EUMAS 2026), 2026, URL.