SPOT: Sensing-augmented Trajectory Planning via Obstacle Threat Modeling

TL;DR

SPOT tackles UAV obstacle avoidance with a single depth camera by unifying perception and trajectory planning. It builds a Gaussian Process-based obstacle belief over both observed and unobserved space, and derives a time-varying observation urgency that guides camera orientation and motion to maximize future perception and safety. The framework optimizes a differentiable, sensing-augmented objective solved in real time with $L$-BFGS, achieving computation times under 10 ms and enabling early detection of dynamic obstacles (e.g., 2.8 s earlier) with substantial gains in visibility. Simulations and real-world experiments demonstrate SPOT’s superiority over decoupled baselines, achieving higher success ratios and longer observation lead times in dynamic, cluttered, and occluded environments.

Abstract

UAVs equipped with a single depth camera encounter significant challenges in dynamic obstacle avoidance due to limited field of view and inevitable blind spots. While active vision strategies that steer onboard cameras have been proposed to expand sensing coverage, most existing methods separate motion planning from sensing considerations, resulting in less effective and delayed obstacle response. To address this limitation, we introduce SPOT (Sensing-augmented Planning via Obstacle Threat modeling), a unified planning framework for observation-aware trajectory planning that explicitly incorporates sensing objectives into motion optimization. At the core of our method is a Gaussian Process-based obstacle belief map, which establishes a unified probabilistic representation of both recognized (previously observed) and potential obstacles. This belief is further processed through a collision-aware inference mechanism that transforms spatial uncertainty and trajectory proximity into a time-varying observation urgency map. By integrating urgency values within the current field of view, we define differentiable objectives that enable real-time, observation-aware trajectory planning with computation times under 10 ms. Simulation and real-world experiments in dynamic, cluttered, and occluded environments show that our method detects potential dynamic obstacles 2.8 seconds earlier than baseline approaches, increasing dynamic obstacle visibility by over 500\%, and enabling safe navigation through cluttered, occluded environments.

Figures (7)

• Figure 1: Comparison between independent planner and coupled planner.
• Figure 2: Overview of the SPOT pipeline.
• Figure 3: Performance Comparison of Different Planning Methods in Different Scenarios: (a) Street scenario. (b) Simple corner scenario. (c) Museum scenario.
• Figure 4: Trajectory comparison in a simple corner environment: (a) Gazebo simulation environment (b) EGO-Planner's trajectory resulting in collision with the obstacle. (c) Our method's trajectory demonstrating successful obstacle avoidance.
• Figure 5: Real-world Experiment and our UAV system. The first row shows the third-person perspective, the second row shows the camera view from the drone, and the simulation map visualization. (a)-(b) depict the indoor corner environment, where the drone avoids suddenly appearing obstacles. (c)-(d) show the outdoor forest environment, where the drone navigates through a forest with multiple dynamic obstacles.