UGV-guided Real-time Path Planning for a Vehicle Platoon in Rough Terrains

Principal Investigators

• Constantinos Chamzas (PI), Worchester Polytechnic Institute
• Jing Xiao, Lee Moradi (co-PI), WPI

Students

• Jaskrit Singh, Abhiroop Ajith, Kashif Noori, WPI

Government

• Philip Frederick, Jon Smereka, U.S. Army GVSC

Industry

• Iain Dodds, VI Grade
• Alyssa Scheske, Applied Intuition

Project #1.A122 begins in 2025.

When planning mission routes for a large platoon of soldier-operated vehicles, it is essential to ensure that these routes are efficient, safe, and traversable. However, the lack of prior knowledge about the terrain or potential enemy locations often hinders the ability to pre-plan such paths effectively. In such cases, a reconnaissance Unmanned Ground Vehicle (UGV) can be deployed to scout unknown sections of the planned route. Utilizing an agile UGV for this task minimizes the risks to human soldiers and protects larger, more expensive vehicles by providing critical real-time information about the environment.

Currently there is no existing planner for designing routes for large platoons that concurrently can satisfy mission objectives (e.g., spatiotemporal constraints) and are feasible for human drivers and vehicles (e.g., human and kinodynamic constraints). Existing research only addresses isolated subtopics of motion planning, focusing only on spatiotemporal constraints or kinodynamic constraints. Second, the challenge of planning reconnaissance motions for a UGV in realtime remains unexplored in the current literature, as it involves online coverage path planning (CPP) and information gathering of unknown or changing environments to facilitate platoon path planning. Most existing CPP planners are offline for known environments. Real-time path planning for robots in changing environments often overlook critical factors such as 3D terrain characteristics. Finally, leveraging both exteroceptive data (e.g., images, 3D sensors) and proprioceptive data (e.g., IMU, wheel encoders) from one vehicle to update the traversability information for another vehicle is also a novel challenge that has not been addressed in the current literature.

Our Research Objectives are:

1. Efficiently generating paths that simultaneously satisfy spatiotemporal and kinodynamic constraints: Develop algorithms to efficiently compute time-parameterized paths for a large platoon of vehicles based on an estimated traversability map and specified geometric/temporal goals (mission objectives). Efficiently handling partial map information to generate multiple potential paths that satisfy the given spatiotemporal and kinodynamic constraints is the primary challenge.
2. Plan and Execute Reconnaissance to Validate Planned Paths: Investigate reconnaissance motion planning for the UGV to verify potential paths for the large vehicle platoon. This involves planning and deploying the UGV to gather data to confirm traversability for both the platoon and itself, ensuring safe navigation along proposed paths.
3. Update Estimated Traversability Map from Heterogeneous Data: Explore updating a traversability map using both exteroceptive (images, 3D sensors) and proprioceptive (IMU, wheel encoders) data. Specifically, we’ll address the challenge of integrating data from a different vehicle, e.g., leveraging data from a UGV to estimate traversability for a platoon of soldier-driven vehicles different from the UGV, which have different kinematic constraints.

#1.A122