Terrain-Aware Model Predictive Control of Heterogeneous Bipedal and Aerial Robot Coordination for Search and Rescue Tasks

TL;DR

A terrain-aware Model Predictive Controller (MPC) that incorporates terrain elevation gradients learned using Gaussian processes (GP) that generates safe navigation paths for the bipedal robots to traverse rough terrain while minimizing terrain slopes, and it directs the quadrotors to perform aerial search and mapping tasks.

Abstract

Humanoid robots offer significant advantages for search and rescue tasks, thanks to their capability to traverse rough terrains and perform transportation tasks. In this study, we present a task and motion planning framework for search and rescue operations using a heterogeneous robot team composed of humanoids and aerial robots. We propose a terrain-aware Model Predictive Controller (MPC) that incorporates terrain elevation gradients learned using Gaussian processes (GP). This terrain-aware MPC generates safe navigation paths for the bipedal robots to traverse rough terrain while minimizing terrain slopes, and it directs the quadrotors to perform aerial search and mapping tasks. The rescue subjects' locations are estimated by a target belief GP, which is updated online during the map exploration. A high-level planner for task allocation is designed by encoding the navigation tasks using syntactically cosafe Linear Temporal Logic (scLTL), and a consensus-based algorithm is designed for task assignment of individual robots. We evaluate the efficacy of our planning framework in simulation in an uncertain environment with various terrains and random rescue subject placements.

Figures (4)

• Figure 1: Overall block diagram of the proposed heterogeneous bipedal and aerial robot planner for search and rescue in terrain-uncertain environments.
• Figure 2: Illustration of the LIP model for bipedal locomotion and the quadrotor model for two consecutive discrete states. The orange lines indicate the range sensors attached to the robots and an illustration of the corresponding local field of view for measuring the terrain data.
• Figure 3: (a) shows the candidate target points $\boldsymbol{t}$ as $\star$, based on the belief GP of the environment shown as the gray-white gradients. (b) shows the sample points for the calculation of the traversability score $\boldsymbol{p}^{\rm ellipse}_i, \; i \in M$, as the black dots inside the ellipse. The ellipse focal points are the current position of the robot and the candidate target point shown as a red $\star$. The terrain GP of the environment is shown by the green-white gradient.
• Figure 4: Planning results: (a) shows the average belief value for two different trials with randomly selected initial belief conditions $\mathcal{B}$. The red line is trial without conflict resolution and the green line is with conflict resolution. (b) shows the lateral slopes traversed by both Digits. The red line is a trial without traversiblity score and slope minimization in the MPC and the green line is with traversiblity score and slope minimization. (c) shows conflict resolution when the auction phase outputs the same targets for the robot team, where the straight lines connect robots to their target points, (d) shows the average standard deviation of the terrain GP, (e) shows the task allocation and assignment results of two different runs with and without wind $O_{E1}$, (f) shows the belief value, and (g) shows that the standard deviation of the terrain estimation decreases as the robots explore the environment.

Theorems & Definitions (6)

• Definition III.1: Gaussian Process Regression
• Definition V.1: Syntactically co-safe LTL belta_formal_2017
• Definition V.2: Finite State Automaton belta_formal_2017
• Definition V.3: Robot-centric Observations
• Definition V.4: Environment-centric Observations
• Definition V.5: Robot Task