Socially Aware Multi-Agent Control using Real-Time STTs

Real-time spatiotemporal tubes for socially aware navigation and multi-agent coordination

An autonomous vehicle delivering groceries is expected to exhibit higher social awareness (altruistic: more cooperative and cautious, prioritizing safety) compared to a fire-truck (egoistic: more goal-driven, prioritizing task completion). The fire truck may move straight ahead, while a commercial vehicle should adjust its behavior to prevent collisions. Thus, social awareness can be viewed as a personality trait, and modeling heterogeneous interactions among agents leads to a more realistic representation of such systems.

Interaction between an egoistic (high-priority) fire truck and an altruistic (collision-avoiding) grocery vehicle.

In this project, we develop a real-time Spatiotemporal Tube (STT) framework that allows socially aware, model-free, and computationally fast control of unknown multi-agent systems.
Each agent is assigned a Social Awareness Index that modulates its behavior: egoistic or altruistic, allowing human-like interactive behavior during navigation.

The approach combines:

  • Real-time STT construction
  • Model-free closed-form control
  • Social interaction functions
  • Multi-agent safety and negotiation
  • Prescribed-time reach-avoid-stay (TRAS) guarantees

To examine the influence of the social awareness index, we considered two scenarios with two omnidirectional robots (blue and yellow). In Case~1, the blue agent was assigned a higher social awareness index than the yellow agent. Consequently, the blue agent behaved more altruistically, showing greater flexibility and cooperation, whereas the yellow agent was more egoistic, prioritizing its goal and obstacle avoidance. In Case~2, the social awareness indices were swapped while keeping all other parameters identical, leading to the opposite behaviors compared to Case~1.

Hardware demonstration of two omnidirectional robots in a cluttered dynamic environment

Related Publication

Incorporating Social Awareness into Control of Unknown Multi-Agent Systems: A Real-Time Spatiotemporal Tubes Approach
R. Das, S. Upadhyay, P. Jagtap
arXiv:2510.25597 (2025)

Read the paper

References