In Amazon robotic warehouses, the destination-to-chute mapping problem is crucial for efficient package sorting. Often, however, this problem is complicated by uncertain and dynamic package induction rates, which can lead to increased package recirculation. To tackle this challenge, we introduce a Distributionally Robust Multi-Agent Reinforcement Learning (DRMARL) framework that learns a destination-to-chute mapping policy that is resilient to adversarial variations in induction rates. Specifically, DRMARL relies on group distributionally robust optimization (DRO) to learn a policy that performs well not only on average but also on each individual subpopulation of induction rates within the group that capture, for example, different seasonality or operation modes of the system. This approach is then combined with a novel contextual bandit-based estimator of the worst-case induction distribution for each state-action pair, significantly reducing the cost of exploration and thereby increasing the learning efficiency and scalability of our framework. Extensive simulations demonstrate that DRMARL achieves robust chute mapping in the presence of varying induction distributions, reducing package recirculation by an average of 80% in the simulation scenario.

Multi-Agent Path Finding (MAPF) is the problem of moving a team of agents to their goal locations without collisions. In this paper, we study the lifelong variant of MAPF, where agents are constantly engaged with new goal locations, such as in large-scale automated warehouses. We propose a new framework Rolling-Horizon Collision Resolution (RHCR) for solving lifelong MAPF by decomposing the problem into a sequence of Windowed MAPF instances, where a Windowed MAPF solver resolves collisions among the paths of the agents only within a bounded time horizon and ignores collisions beyond it. RHCR is particularly well suited to generating pliable plans that adapt to continually arriving new goal locations. We empirically evaluate RHCR with a variety of MAPF solvers and show that it can produce high-quality solutions for up to 1,000 agents (= 38.9% of the empty cells on the map) for simulated warehouse instances, significantly outperforming existing work.

Autonomous mobile robots need maps for effective, safe navigation, and SLAM in general is still an unsolved problem. Nonetheless, certain combinations of environmental characteristics and sensors admit tractable solutions. In particular, detection and tracking of linear features such as line segments (2D) or planar facets (3D) has been proven robust in many man-made environments. However, these types of features produce rank-deficient constraints, which create challenges for graph-based SLAM optimizers. We present techniques for using rank-deficient features and constraints more robustly by analyzing the approximate null-space of the constraints for each node in the factor graph representing the trajectory. We also extend auxiliary methods for correspondence calculations and map update routines, the combination of which yields state-of-the-art performance for a rank-deficient SLAM system. We present results from quantitative experiments comparing memory use, compute load, accuracy, and robustness for several ablation tests on real and simulated data.

Robotic challenges like the Amazon Picking Challenge (APC) or the DARPA Challenges are an established and important way to drive scientific progress. They make research comparable on a well-defined benchmark with equal test conditions for all participants. However, such challenge events occur only occasionally, are limited to a small number of contestants, and the test conditions are very difficult to replicate after the main event. We present a new physical benchmark challenge for robotic picking: the ACRV Picking Benchmark. Designed to be reproducible, it consists of a set of 42 common objects, a widely available shelf, and exact guidelines for object arrangement using stencils. A well-defined evaluation protocol enables the comparison of complete robotic systems - including perception and manipulation - instead of sub-systems only. Our paper also describes and reports results achieved by an open baseline system based on a Baxter robot.

We propose distributed algorithms to automatically deploy a team of mobile robots to partition and provide coverage of a nonconvex environment. To handle arbitrary nonconvex environments, we represent them as graphs. Our partitioning and coverage algorithm requires only short-range, unreliable pairwise “gossip” communication. The algorithm has two components: 1) a motion protocol to ensure that neighboring robots communicate at least sporadically and 2) a pairwise partitioning rule to update territory ownership when two robots communicate. By studying an appropriate dynamical system on the space of partitions of the graph vertices, we prove that territory ownership converges to a pairwise-optimal partition in finite time. This new equilibrium set represents improved performance over common Lloyd-type algorithms. Additionally, our algorithm is an "anytime algorithm'' that also scales well for large teams and can be run by on-board computers with limited resources. Finally, we report on large-scale simulations in complex environments and hardware experiments using the Player/Stage robot control system.

This paper focuses on the problem of patrolling an environment with a team of autonomous agents. Given a set of strategically important locations (viewpoints) with different priorities, our patrolling strategy consists of 1) constructing a tour through the viewpoints, and 2) driving the robots along the tour in a coordinated way. As performance criteria, we consider the weighted refresh time, i.e., the longest time interval between any two visits of a viewpoint, weighted by the viewpoint's priority. We consider the design of both optimal trajectories and distributed control laws for the robots to converge to optimal trajectories. First, we propose a patrolling strategy and we characterize its performance as a function of the environment and the viewpoints priorities. Second, we restrict our attention to the problem of patrolling a nonintersecting tour, and we describe a team trajectory with minimum weighted refresh time. Third, for the tour patrolling problem and for two distinct communication scenarios, namely the Passing and the Neighbor-Broadcast communication models, we develop distributed algorithms to steer the robots toward a minimum weighted refresh time team trajectory. Finally, we show the effectiveness and robustness of our control algorithms via simulations and experiments.

This paper addresses a novel visibility-based pursuit-evasion problem in which a team of searchers with limited range sensors must coordinate to clear any evaders from an unknown planar environment. We present a distributed algorithm built around guaranteeing complete coverage of the frontier between cleared and contaminated areas while expanding the cleared area. Our frontier-based algorithm can guarantee detection of evaders in unknown, multiply-connected planar environments which may be non-polygonal. We also detail a method for storing and updating the global frontier between cleared and contaminated areas without building a global map or requiring global localization, which enables our algorithm to be truly distributed. We demonstrate the functionality of the algorithm through Player/Stage simulations.

This paper presents a new method for reactive collision avoidance for mobile robots in complex and cluttered environments. Our technique is to adapt the ldquodivide and conquerrdquo approach of the nearness-diagram+ navigation (ND+) method to generate a single motion law which applies for all navigational situations. The resulting local path planner considers all the visible obstacles surrounding the robot, not just the closest two. With these changes our new navigation method generates smoother motion while avoiding obstacles. Results from comparisons with ND+ are presented as are experiments using Erratic mobile robots.