CoMRoS: Cooperative Mobile Robots Stuttgart

AAAI 1996

TL;DR

This paper introduces CoMRoS, a system for cooperative mobile robots that aims to achieve complex tasks autonomously through decentralized plan exchange. The system utilizes off-the-shelf "Robuter II" robots and demonstrates its capabilities in two robot competition scenarios: "call a meeting" and "clean up the tennis court."

What Problem Does This Paper Address?

The paper addresses the challenge of enabling multiple mobile robots to collaborate and achieve a common goal without a central controller. Traditional approaches often rely on a single point of control, which can be a bottleneck and a single point of failure. CoMRoS aims to overcome this by allowing robots to exchange plans and coordinate their actions autonomously, making the system more robust and scalable. This is particularly relevant for tasks requiring complex coordination, such as those found in robot competitions.

Methods

The CoMRoS system uses "Robuter II" vehicles from Robosoft France, which are equipped with a VME bus system and controlled remotely via wireless Ethernet for steering commands and sonar data. A wireless video link transmits camera images. Data exchange between robots occurs through corresponding workstations. For testing and debugging, remote control is used, but the system is designed for complete autonomous operation using a laptop PC. The robots are programmed using the Linux operating system on the PC, while the robots themselves run a real-time operating system. The paper mentions the implementation of libraries for Ethernet communication, sonar data polling, and video frame digitization. For specific tasks, the system employs a Floyd algorithm for shortest path determination and a simple vision difference algorithm for room occupancy detection. For the "clean up the tennis court" task, a rotating cylinder was developed for ball collection, with subsequent image processing to identify missed balls.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Decentralized Control for Complex Tasks: The core idea is that robots can achieve sophisticated tasks by sharing plans and coordinating locally, rather than relying on a central commander. This enhances robustness and scalability.
  • Leveraging Off-the-Shelf Hardware: The project demonstrates that even with relatively standard mobile robot platforms ("Robuter II"), complex cooperative behaviors can be implemented.
  • Hybrid Control Strategy: The system supports both remote control for development and full autonomy for deployment, offering flexibility in the development lifecycle.
  • Task-Specific Adaptations: The paper highlights how the general CoMRoS framework can be adapted for different competition tasks, showcasing the system's versatility. For instance, the "clean up the tennis court" task evolved from a manipulator-based approach to a simpler harvester-like mechanism.

Key Results

The paper describes the configuration of one robot for autonomous operation in two competition events:

  • Event 1: "Call a meeting": This task relies heavily on a 24-sensor sonar array for navigation and obstacle detection. The system also explores using a vision difference algorithm to detect if a conference room is empty and potentially voice input.
  • Event 2: "Clean up the tennis court": This task involves a rotating cylinder mounted on the robot to collect balls. The collection process is described as an "area filling algorithm," with balls unloaded by reversing the cylinder's rotation. Subsequent image processing is used to find missed balls.

The paper does not present quantitative results in terms of accuracy or performance metrics for these tasks.

Limitations and Future Work

The paper acknowledges that the initial approach for the "clean up the tennis court" task, which involved a manipulator, was too time-consuming due to image processing and control complexity. The paper suggests that for the "call a meeting" task, implementing voice input is an alternative being investigated. Further details on the performance and robustness of the autonomous systems are not provided, which could be a direction for future work.

Summary updated

BibTeX
@inproceedings{1996comros,
  title = {{CoMRoS: Cooperative Mobile Robots Stuttgart}},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year = {1996},
}
Figure