Multi-robot navigation is a challenging task in which multiple robots must be coordinated simultaneously within dynamic environments. We apply Deep Reinforcement Learning (DRL) to learn a decentralized end-to-end policy which maps raw sensor data to the command velocities of the agent. In order to enable the policy to generalize, the training is performed in different environments and scenarios. The trained policy is tested and evaluated in common multi-robot scenarios like switching a place, an intersection and a bottleneck situation. The learned policy allows the agent to recover from dead ends and to navigate through complex environments.
The measurement of the growth state and health status of single plants or even single parts of the plants within a crop to conduct precision farming actions is a difficult task. We address this challenge by adopting a multi-sensor suite, which can be used on several sensor-platforms. Based on experimental field studies in relevant agricultural environments, we show how the acquired hyperspectral, LIDAR, stereo and thermal image data can be processed and classified to get a comprehensive understanding of the agricultural acreage.
ecent successes aside, reinforcement learning (RL) still faces significant challenges in its application to the real-world robotics domain. Guiding the learning process with additional knowledge offers a potential solution, thus leveraging the strengths of data-and knowledge-driven approaches. However, this field of research encompasses several disciplines and hence would benefit from a structured overview.In this article, we propose a concept of guided RL that provides a systematic approach toward accelerating the training process and improving performance for real-world robotics settings. We introduce a taxonomy that structures guided RL approaches and shows how different sources of knowledge can be integrated into the learning pipeline in a practical way. Based on this, we describe available approaches in this field and quantitatively evaluate their specific impact in terms of efficiency, effectiveness, and sim-to-real transfer within the robotics domain. However, learning control policies in such a way naturally requires many interactions with the environment. This emphasizes the importance of both collecting highquality samples and exploring the search space in a sample-efficient manner. While directly learning on real robots is appealing, it comes along with substantial challenges, such as high sample cost, partial observability, and safety constraints [28]. Hence, simulators are often
Autonomous Mobile Robots (AMRs) continue to facilitate the work of physicians and hospital nurses by releasing those professionals from time consuming transport tasks within hospitals. Nonetheless, AMRs still often face challenges when situations occur, which result in a failure of the navigation system. In this paper, we present an analysis and an implementation of a remote-control mechanism using 5G networks to enable an operator to control an AMR, in our example within a hospital, to support an AMR in situations, where an autonomous navigation faced challenges, that cannot be solved autonomously. In detail, four major challenges are faced when implementing a remote control for failure situations – the data connection itself, the sensor data acquisition and compression, the delivery of the current robot state for a user and the controllability of the robot. For autonomous driving, the AMR is equipped with a 128-layered 3D-Lidar sensor. An RGB-D camera facilitates video feedback for the operator to navigate the AMR manually. Additionally, the point cloud of the 3D-lidar provides a detailed in-depth view of the environment, which recognizes present persons or also allows the operator to drive backwards. To establish a connection between an AMR and a remote operator, a stable and low latency data connection is required. Since the Wi-Fi requirements of hospitals usually do not fit the requirements of remote-controlled robots regarding data security, network coverage, connection latency and bandwidth, the usage of the Wi-Fi network of hospitals is not appropriate. These challenges can be overcome using 5G cellular network to guarantee a low latency, high bandwidth connection which is independent from the regulations and limitations of the local Wi-Fi network. Nonetheless, by selecting the cellular 5G network as the remote operation network, further challenges arise – e.g. coverage of the 5G network or the stable and secure accessibility of the robot. Since hospital building structures are complex and usually are constructed using reinforced concrete, 5G radio waves are reflected or absorbed. In addition, the bandwidth is limited, since a public cellular connection is used. Due to these limitations, data compression is required for transmitting large chunks of sensor data, such as RGB camera streams or point clouds. The RGB video compression is implemented using the H.264 codec, which again can be accelerated using hardware. The point cloud is compressed through an octree implementation. As a result, the sensor data is transmitted with low latency and less lag. Despite using data compression algorithms, which are not lossless, the quality of the sensor data, received by the operator, is still sufficient for remote control operations. For a safe and controlled remote control of an AMR using the above explained technology stack, a data connection with less to no data transfer loss is required.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.