The advent of tactile sensors in robotics has sparked many ideas on how robots can leverage direct contact measurements of their environment interactions to improve manipulation tasks. An important line of research in...
详细信息
ISBN:
(纸本)9798350377712;9798350377705
The advent of tactile sensors in robotics has sparked many ideas on how robots can leverage direct contact measurements of their environment interactions to improve manipulation tasks. An important line of research in this regard is grasp force control, which aims to manipulate objects safely by limiting the amount of force exerted on the object. While prior works have either hand-modeled their force controllers, employed model-based approaches, or not shown sim-to-real transfer, we propose a model-free deep reinforcement learning approach trained in simulation and then transferred to the robot without further fine-tuning. We, therefore, present a simulation environment that produces realistic normal forces, which we use to train continuous force control policies. A detailed evaluation shows that the learned policy performs similarly or better than a hand-crafted baseline. Ablation studies prove that the proposed inductive bias and domain randomization facilitate sim-to-real transfer. Code, models, and supplementary videos are available on https://***/view/rl-force-ctrl
A number of guidelines, policies, tactics, and procedures served as the foundation for safety in the rail transport system. The personnel's professional abilities must be trained and maintained, but the most cruci...
详细信息
A number of guidelines, policies, tactics, and procedures served as the foundation for safety in the rail transport system. The personnel's professional abilities must be trained and maintained, but the most crucial step in putting them into practice is having them interconnect logic schemes initially created with mechanical installations and subsequently logic schemes involving relays. It was necessary to update the regulations and modernize the signaling systems due to the events and mishaps that occurred on the railway network and the growth in traffic. Using the Cyber-Physical System (CPS) approach may be the most effective way to introduce new security techniques. A significant area that has been examined by other writers in earlier projects is the thermal inspections of the train and infrastructure components using certain sensors that are linked to the machinery used for railway traffic control. Because of the basic concepts of CPS, the typical layered structure, the large computing and communication capabilities, and the similarity to conventional controlsystems, the progress and evolution of the railway system must use and apply CPS components. Many factors are involved in this action, from the large rolling stock manufacturers to the particular users. In this work the possibilities to implement the CPS concepts and the potential advantages in the railway system are emphasized. Besides thermal sensors, sensors for other variables and the possibility of integrating them in an intelligent system are studied. On the other hand, very important in the CPS concept are the models for equipment, operation, maintenance, simple example being given.
In this article, an intelligent embedded electromobility is designed, which combines edge safety, array headlamp module as the optical fusion, vision fusion, and then successfully experimental on an electromobility. T...
详细信息
ISBN:
(纸本)9798350359893;9798350359886
In this article, an intelligent embedded electromobility is designed, which combines edge safety, array headlamp module as the optical fusion, vision fusion, and then successfully experimental on an electromobility. The edge safety of embedded electromobility is based on the image recognition technology for detecting surrounding electromobility and lane lines, and further edge computing as the basis of computer vision within embedded electromobility. Thus, this article can implement two developed, such as lane following and lane deviation warning system by detection algorithm. The proposed array headlamp module promoted safety distance is experimental. Finally, the presented integrated fusion has been incorporated into an electrical vehicle and achieved in a real-road condition, where the artificial intelligence engine of the whole sensors, actuators as well as vehicle control unit (VCU) are used the intelligence of thing (IoT) controller area network (CAN) bus to realize control and signals collection.
Manipulating robots using natural language is the preferred way for non-technical specialists. The challenge lies in reliability and adaptability especially when robots operate in unstructured surroundings. In this pa...
详细信息
ISBN:
(纸本)9798350377712;9798350377705
Manipulating robots using natural language is the preferred way for non-technical specialists. The challenge lies in reliability and adaptability especially when robots operate in unstructured surroundings. In this paper, we propose a novel framework called Dialogue Generative Behavior Trees (DiaGBT). Natural language instructions from human operators are transformed into behavior trees (BTs) and further executed by robots. Compared to the emerging Large Language Models (LLMs), DiaGBT is comparable in terms of semantic understanding but more lightweight, since the parsing rules are produced by LLM but tailored for task-correlated instructions. Besides, DiaGBT allows multi-round human-robot interaction, where robots learn reusable skills in real time. For evaluation, we generate a dataset with 4k instruction-BT pairs covering 4 different scenarios. On average, DiaGBT reaches over 90% parsability and 80% plausibility. Similar results on the VEIL-500 dataset outperform the current state of the art.
In recent years, the field of I'CT education has emerged, and the research and development of educational support robots has attracted significant interest. In this study, we focus on a perplexity estimation metho...
详细信息
The data center can cooperate with the integrated energy system in the park, to improve the flexibility of load regulation. Aiming at the conflict of interest caused by the cooperation among the integrated energy syst...
详细信息
In the last decade, data-driven approaches have become popular choices for quadrotor control, thanks to their ability to facilitate the adaptation to unknown or uncertain flight conditions. Among the different data-dr...
详细信息
ISBN:
(纸本)9798350377712;9798350377705
In the last decade, data-driven approaches have become popular choices for quadrotor control, thanks to their ability to facilitate the adaptation to unknown or uncertain flight conditions. Among the different data-driven paradigms, Deep Reinforcement Learning (DRL) is currently one of the most explored. However, the design of DRL agents for Micro Aerial Vehicles (MAVs) remains an open challenge. While some works have studied the output configuration of these agents (i.e., what kind of control to compute), there is no general consensus on the type of input data these approaches should employ. Multiple works simply provide the DRL agent with full state information, without questioning if this might be redundant and unnecessarily complicate the learning process, or pose superfluous constraints on the availability of such information in real platforms. In this work, we provide an in-depth benchmark analysis of different configurations of the observation space. We optimize multiple DRL agents in simulated environments with different input choices and study their robustness and their sim-to-real transfer capabilities with zero-shot adaptation. We believe that the outcomes and discussions presented in this work supported by extensive experimental results could be an important milestone in guiding future research on the development of DRL agents for aerial robot tasks.
In the context of a Human-in-the-Loop (HITL) system, the accuracy of reachability analysis plays a significant role in ensuring the safety and reliability of HITL systems. In addition, one can avoid unnecessary conser...
详细信息
ISBN:
(纸本)9798350355376;9798350355369
In the context of a Human-in-the-Loop (HITL) system, the accuracy of reachability analysis plays a significant role in ensuring the safety and reliability of HITL systems. In addition, one can avoid unnecessary conservativeness by explicitly considering human control behavior compared to those methods that rely on the system dynamics alone. One possible approach is to use a Gaussian Mixture Model (GMM) to encode human control behavior using the Expectation-Maximization (EM) algorithm. However, relatively few works consider the admissible control input ranges due to physical or mechanical limitations when modeling human control behavior. This could make the following reachability analysis overestimate the system's capability, thereby affecting the performance of the HITL system. To address this issue, we present a constrained stochastic reachability analysis algorithm that can explicitly account for the admissible control input ranges. By confining the ellipsoidal confidence region of each Gaussian component using Sequential Quadratic Programming (SQP), we probabilistically constrain the GMM as well as the corresponding stochastic reachable sets. A comprehensive mathematical analysis of how the constrained GMM can affect the stochastic reachable sets is provided in this paper. Finally, the proposed stochastic reachability analysis algorithm is validated via an illustrative numerical example.
Today, intelligent robotic manufacturing systems are reshaping the production industry. Using robots as actuators, multi-source sensors for perception, and Artificial Intelligence (AI) as decision-making systems, they...
详细信息
ISBN:
(纸本)9798350377712;9798350377705
Today, intelligent robotic manufacturing systems are reshaping the production industry. Using robots as actuators, multi-source sensors for perception, and Artificial Intelligence (AI) as decision-making systems, they can perform routine manufacturing tasks, surpassing the capabilities of traditional hard-programmed Computer Numerical control (CNC) machinery. One specific challenge in footwear manufacturing is sole deburring, traditionally done manually by skilled workers. This paper focuses on developing a robust path-planning pipeline, comprising vision-based and Learning from Demonstrations (LfD) modules for autonomous deburring of soles. The vision-based module exploits Deep Learning (DL) techniques to handle key challenges such as precise segmentation of different soles types across diverse scenarios despite potential occlusions. Additionally, a novel method for burrs identification has been developed leveraging image processing and optimization techniques. Determining the optimal cutting tool orientation during sole deburring relies on human experience. The LfD module aims to impart this knowledge to the robot from videos of expert demonstrations, requiring adaptability to every new incoming sole that needs deburring. Experimental results showcase the method's performance and flexibility, underlining the potential to advance the field of the proposed approach.
Under the background of the 'dual carbon' goal and related policies and the construction of new power systems, the large-scale deployment of sensing terminals and the explosive growth of the perception amount ...
详细信息
暂无评论