IA-TIGRIS: An Incremental and Adaptive Sampling-Based Planner for Online Informative Path Planning

Abstract

Planning paths that maximize information gain for robotic platforms has wide-ranging applications and significant potential impact. To effectively adapt to real-time data collection, informative path planning must be computed online and be responsive to new observations. In this work, we present IA-TIGRIS (Incremental and Adaptive Tree-based Information Gathering Using Informed Sampling), which is an incremental and adaptive sampling-based informative path planner designed for real-time onboard execution. Our approach leverages past planning efforts through incremental refinement while continuously adapting to updated belief maps. We additionally present detailed implementation and optimization insights to facilitate real-world deployment, along with an array of reward functions tailored to specific missions and behaviors. Extensive simulation results demonstrate IA-TIGRIS generates higher-quality paths compared to baseline methods. We validate our planner on two distinct hardware platforms: a hexarotor unmanned aerial vehicle (UAV) and a fixed-wing UAV, each having different motion models and configuration spaces. Our results show up to a 38% improvement in information gain compared to baseline methods, highlighting the planner's potential for deployment in real-world applications. Project website: https://ia-tigris.github.io

Figures (19)

• Figure 1: IA-TIGRIS deployed on a hexarotor UAV to map the location of cars in the environment. The visualization shows a probability grid representing the prior belief of car locations and a representative path generated by our planner. The algorithm runs entirely onboard, continuously refining and adapting paths based on updated world beliefs from sensor observations.
• Figure 2: The IA-TIGRIS planning framework consists of two main components: the belief map and the informative path planner. The belief map is responsible for managing the agent's understanding of the environment while the informative path planner generates adaptive plans based on the agent's position and remaining budget. For the planning procedure, the new path is merged with the existing plan by pruning outdated portions of the existing path and starting the new plan from the agent's expected position.
• Figure 3: Visualization of the online replanning procedure. The future robot state is used as the starting point for the next plan. The newly-planned path is merged with the previous path.
• Figure 4: An illustration of the incremental planning and refinement process in IA-TIGRIS. (a) The global planner first generates long-horizon paths to maximize predicted information gain. The blue ellipses correspond to areas of information where areas with a larger alpha correspond to higher information. (b) At the start of each new planning cycle, infeasible portions of the tree, based on the executed trajectory and the current plan, are pruned. The remaining tree is then updated by recomputing information gain and cost based on the latest belief state and available budget, which is significantly more efficient than rebuilding the tree from scratch. (c) The tree is further expanded and refined for the duration of the planning cycle, after which the process repeats.
• Figure 5: An example of how the grid cells rewards are estimated given a planned trajectory. The cells that are closer to the center of the trajectory have a smaller minimum distance than cells farther from the center of the trajectory, leading to a larger change in entropy for the closer cells.