A Fleet of Miniature Cars for Experiments in Cooperative Driving (1902.06133v1)

Abstract: We introduce a unique experimental testbed that consists of a fleet of 16 miniature Ackermann-steering vehicles. We are motivated by a lack of available low-cost platforms to support research and education in multi-car navigation and trajectory planning. This article elaborates the design of our miniature robotic car, the Cambridge Minicar, as well as the fleet's control architecture. Our experimental testbed allows us to implement state-of-the-art driver models as well as autonomous control strategies, and test their validity in a real, physical multi-lane setup. Through experiments on our miniature highway, we are able to tangibly demonstrate the benefits of cooperative driving on multi-lane road topographies. Our setup paves the way for indoor large-fleet experimental research.

• The paper introduces a low-cost testbed featuring 16 miniature Ackermann-steering vehicles for cooperative driving experiments.
• It employs real-time motion capture and decentralized control to achieve nonlinear trajectory tracking with a 35-42% traffic throughput increase.
• The setup enables practical research in human-like and autonomous driving models, paving the way for innovative traffic management systems.

Overview of 'A Fleet of Miniature Cars for Experiments in Cooperative Driving'

The paper, 'A Fleet of Miniature Cars for Experiments in Cooperative Driving', presents a low-cost experimental testbed designed to advance research and education in cooperative driving strategies using miniature Ackermann-steering vehicles. Developed at the University of Cambridge, this setup addresses the challenges and expenses associated with larger-scale facilities, facilitating indoor experiments on cooperative driving with practical implications for real-world auto systems.

The Cambridge Minicar platform comprises 16 inexpensive, downsized robotic vehicles equipped with fundamental vehicle dynamics akin to real cars. The testbed uses decentralized control strategies implemented through both state-of-the-art driver models and autonomous algorithms. The system incorporates these vehicles in a multi-lane, miniature freeway environment, aiming to validate various cooperative and competitive driving strategies.

Key Contributions

• Design and Cost Efficiency: The Cambridge Minicar is a standout feature due to its low cost (approximately USD 76 per unit), which allows for large fleet assembly. It provides the ability to test substantial vehicle-to-vehicle interactions, crucial for studying traffic scenarios.
• Testbed Architecture: The system leverages a motion capture system for real-time data acquisition and trajectory tracking with positional accuracy. This is complemented by a simplified computational approach that retains resource-heavy computations offboard, enhancing setup efficiency without compromising the scalability.
• Trajectory Tracking Implementation: Leveraging a bicycle model with Ackermann steering geometry, the paper details nonlinear trajectory tracking strategies to ensure real-time vehicle operation within the testbed.
• Driving Behavior Models: The use of decentralized strategies emulates human-like driving via IDM and MOBIL models while facilitating cooperative behaviors through virtual vehicle projections and enhanced IDM and MOBIL controls.

Experimental Insights

Experiments demonstrate the system's capability in handling both normal and aggressive driving strategies within egocentric and cooperative contexts. Results quantify the efficacy of cooperative driving approaches, indicating a 35-42% increase in traffic throughput compared to egocentric models. Moreover, gamification allows human interaction within the testbed, offering practical insights into traffic dynamics and algorithm robustness.

Implications and Future Research Directions

This platform stands to significantly impact the development of cooperative driving technologies, potentially influencing traffic management systems and autonomous vehicle controls by providing a readily accessible testbed for algorithm validation and development. Future research directions may include:

• Complex road topographies and intersection management in a dense traffic setup.
• Analysis of heterogeneous behaviors, incorporating both cooperative and non-cooperative models.
• Investigation into the influence of noise and delayed communication on vehicle control strategies.
• Optimization methodologies that include comfort-based metrics, considering driver comfort as a primary factor in vehicle acceleration management.

The paper provides a robust framework for analyzing cooperative driving mechanisms in controlled environments, paving the way for substantive advancements in autonomous vehicle research and real-world traffic management system innovations.