Temporally-Sampled Efficiently Adaptive State Lattices for Autonomous Ground Robot Navigation in Partially Observed Environments

TL;DR

TSEASL is a regional planner arbitration architecture that considers updated and optimized versions of previously generated trajectories against the currently generated trajectory, and results indicate that when running TSEASL, the robot did not require manual interventions in the same locations where the robot was running the baseline planner.

Abstract

Due to sensor limitations, environments that off-road mobile robots operate in are often only partially observable. As the robots move throughout the environment and towards their goal, the optimal route is continuously revised as the sensors perceive new information. In traditional autonomous navigation architectures, a regional motion planner will consume the environment map and output a trajectory for the local motion planner to use as a reference. Due to the continuous revision of the regional plan guidance as a result of changing map information, the reference trajectories which are passed down to the local planner can differ significantly across sequential planning cycles. This rapidly changing guidance can result in unsafe navigation behavior, often requiring manual safety interventions during autonomous traversals in off-road environments. To remedy this problem, we propose Temporally-Sampled Efficiently Adaptive State Lattices (TSEASL), which is a regional planner arbitration architecture that considers updated and optimized versions of previously generated trajectories against the currently generated trajectory. When tested on a Clearpath Robotics Warthog Unmanned Ground Vehicle as well as real map data collected from the Warthog, results indicate that when running TSEASL, the robot did not require manual interventions in the same locations where the robot was running the baseline planner. Additionally, higher levels of planner stability were recorded with TSEASL over the baseline. The paper concludes with a discussion of further improvements to TSEASL in order to make it more generalizable to various off-road autonomy scenarios.

Figures (8)

• Figure 1: The proposed TSEASL architecture for regional planning.
• Figure 2: An illustration of the search space explored by ARA* in a trajectory-aligned state lattice used in the $t_{prev}'$ optimizer step of the architecture proposed in Figure \ref{['fig:tseasl_flowchart']}. The map was synthetically generated using Perlin noise perlin2002improving.
• Figure 3: The forested environment TSEASL was tested in to compare its performance against the baseline.
• Figure 4: Two consecutive planning cycles illustrating when $t_{now}$ has a sufficiently cheaper cost than $t_{prev}'$. The cost of $t_{prev}'$ was 56.83 seconds, while the cost of $t_{now}$ in the right figure was 52.87 seconds. Since this cost difference is $>$ 5% (approximately 6.98%), TSEASL decides that it is worth it to switch its $t_{sel}$ to pass down to the local planner. The orange borders indicate that this example took the orange decision line in Figure \ref{['fig:tseasl_flowchart']}.
• Figure 5: Three consecutive planning cycles when $t_{now}$ does not have a sufficiently cheaper cost than $t_{prev}'$. The cost of $t_{now}$ changed from 48.24 seconds, to 48.28 seconds, back to 48.24 seconds, while the cost of $t_{prev}'$ in all three figures stayed at 48.70 seconds. The cost difference between $t_{now}$ and $t_{prev}'$ was never $>$ 5%, so $t_{sel}$ never changes. The red borders mean this example took the red decision line in Figure \ref{['fig:tseasl_flowchart']}.