PREVENT-JACK: Context Steering for Swarms of Long Heavy Articulated Vehicles
New algorithm prevents jackknifing in swarms of up to 10-trailer vehicles
A team from the University of Magdeburg (Adrian Baruck, Michael Dubé, Christoph Steup, Sanaz Mostaghim) has published PREVENT-JACK, a novel context steering framework designed for swarms of long Heavy Articulated Vehicles (HAVs). Unlike traditional swarm robotics research that treats robots as point masses, this work addresses the unique challenges of kinematically constrained, elongated, and articulated vehicles. The system fuses six local behaviors to provide guarantees against jackknifing and collisions, tested for vehicles with up to ten trailers.
The researchers used 15,000 simulations to evaluate swarm performance, highlighting the importance of an 'Evade Attraction' behavior for deadlock prevention. Results showed that dead- and livelocks occur more frequently in larger swarms and denser scenarios, affecting a peak average of 27% and 31% of vehicles respectively. Larger swarms exhibited increased waiting behaviors, while smaller swarms showed more evasion. The paper is submitted to the Nature Portfolio Journal Robotics (NPJ Robot) and spans 32 pages with 7 figures.
- PREVENT-JACK fuses six local behaviors to prevent jackknifing in swarms of articulated vehicles with up to ten trailers
- 15,000 simulations showed dead- and livelocks affect 27%/31% of vehicles in dense scenarios
- Larger swarms exhibit increased waiting behaviors while smaller swarms show more evasion
Why It Matters
Enables safe, decentralized coordination of real-world heavy vehicle swarms like truck convoys and industrial transport