Optimization of Approximate Clearance Evaluation [ACE] Algorithm for M2020 rover

NASA’s M2020 Perseverance rover faces significant challenges while navigating the dense, uneven terrain of the Jezero Crater, which demands advanced real-time path-planning capabilities [1] Several onboard algorithms and systems on the rover are responsible for ensuring safe traversal across the terrain, namely the Approximate Clearance Evaluation algorithm (ACE), and the navigation stack: ENAV (Enhanced Navigation). ACE and ENAV approach path planning by generating conservative paths based on bounding approximations

To enhance the efficiency of the onboard ACE algorithm, I conducted workload performance analysis on a kinematic model of the Perseverance Rover and designed a high-performance hardware-software implementation for rover path planning. This implementation integrates one or more hardware-accelerated kernels to optimize computational performance and real-time execution.

 Objective  

Reverse-Engineering NASAs ACE Algorithm

NASA Perseverance Rover on ROS/Gazebo

NASA’s Mars 2020 rover mission setting in the Jezero crater has a significantly higher density of rocky obstacles (measured as CFA, i.e., the cumulative fraction of area covered by rocks) than previous Mars missions: up to an estimated 15%-20% CFA, versus CFA under 10% on earlier missions.

Development of Kinematic & Algorithmic Digital Twin

The overall ACE algorithm cost [1] mainly comprises four components discussed in this section. To ensure the proper stability of the rover is maintained, four different types of costs are associated with the overall planner. This section discusses the related costs as follows:

[1]. Otsu K, Matheron G, Ghosh S, Toupet O, Ono M. Fast approximate clearance evaluation for rovers with articulated suspension systems. J Field Robotics. 2020;37:768–785. https://doi.org/10.1002/rob.21892