检索结果-内蒙古大学图书馆

Entropy-Guided Reinforced Open World Active 3D Object Detection Learning

学校读者我要写书评

暂无评论

Entropy-Guided Reinforced Open World Active 3D Object Detect...

2024 China Automation Congress, CAC 2024

作者： Zhang, Haozhe Ma, Liyan Ying, Shihui Institute of Artificial Intelligence Shanghai University Shanghai200444 China School of Computer Engineering and Science School of Mechatronic Engineering and Automation Shanghai Key Laboratory of Intelligent Manufacturing and Robotics Shanghai University Shanghai200444 China Department of Mathematics School of Science Shanghai University Shanghai200444 China

ISBN: (纸本)9798350368604

Traditional fully annotated closed set 3D object detection methods improve model performance but are impractical in real-world settings due to the emergence of new categories and the complexity of 3D annotations. Open-World Object Detection (OWOD) addresses these issues but relies heavily on manual labeling, which is costly. This paper focuses on open world active learning and proposes an entropy-guided reinforced open world active 3D object detection (EROA). EROA regards active learning as a reinforcement learning problem tailored for open driving scenarios. We use entropy as a reward metric for efficient reinforcement learning. We also leverage knowledge from the 2D domain using object-level large-scale vision-language models to enhance sample selection. Extensive experiments evidence that the proposed EROA meets the dynamic and cost-sensitive requirements of autonomous driving, enabling real-time detection of both known and unknown objects. © 2024 IEEE.

关键词： Active learning

Toward a Low-Cost Perception System in Autonomous Vehicles: A Spectrum Learning Approach

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Alsakabi, Mohammed Erickson, Aidan Dolan, John M. Tonguz, Ozan K. Department of Electrical and Computer Engineering College of Engineering The Robotics Institute School of Computer Science Carnegie Mellon University PittsburghPA United States

We present a cost-effective new approach for generating denser depth maps for Autonomous Driving (AD) and Autonomous Vehicles (AVs) by integrating the images obtained from deep neural network (DNN) 4D radar detectors with conventional camera RGB images. Our approach introduces a novel pixel positional encoding algorithm inspired by Bartlett’s spatial spectrum estimation technique. This algorithm transforms both radar depth maps and RGB images into a unified pixel image subspace called the Spatial Spectrum, facilitating effective learning based on their similarities and differences. Our method effectively leverages high-resolution camera images to train radar depth map generative models, addressing the limitations of conventional radar detectors in complex vehicular environments, thus sharpening the radar output. We develop spectrum estimation algorithms tailored for radar depth maps and RGB images, a comprehensive training framework for data-driven generative models, and a camera-radar deployment scheme for AV operation. Our results demonstrate that our approach also outperforms the state-of-the-art (SOTA) by 27.95% in terms of Unidirectional Chamfer Distance (UCD). © 2025, CC BY-NC-ND.

关键词： Radar imaging

Multi-Agent Path Finding Using Conflict-Based Search and Structural-Semantic Topometric Maps

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Fredriksson, Scott Bai, Yifan Saradagi, Akshit Nikolakopoulos, George Robotics and AI group Department of Computer Science Electrical and Space Engineering Luleå University of Technology Sweden

As industries increasingly adopt large robotic fleets, there is a pressing need for computationally efficient, practical, and optimal conflict-free path planning for multiple robots. Conflict-Based Search (CBS) is a popular method for multi-agent path finding (MAPF) due to its completeness and optimality;however, it is often impractical for real-world applications, as it is computationally intensive to solve and relies on assumptions about agents and operating environments that are difficult to realize. This article proposes a solution to overcome computational challenges and practicality issues of CBS by utilizing structural-semantic topometric maps. Instead of running CBS over large grid-based maps, the proposed solution runs CBS over a sparse topometric map containing structural-semantic cells representing intersections, pathways, and dead ends. This approach significantly accelerates the MAPF process and reduces the number of conflict resolutions handled by CBS while operating in continuous time. In the proposed method, robots are assigned time ranges to move between topometric regions, departing from the traditional CBS assumption that a robot can move to any connected cell in a single time step. The approach is validated through real-world multi-robot path-finding experiments and benchmarking simulations. The results demonstrate that the proposed MAPF method can be applied to real-world non-holonomic robots and yields significant improvement in computational efficiency compared to traditional CBS methods while improving conflict detection and resolution in cases of corridor symmetries. Copyright © 2025, The Authors. All rights reserved.

关键词： Continuous time systems

A Graph-Based Reinforcement Learning Approach with Frontier Potential Based Reward for Safe Cluttered Environment Exploration

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Calzolari, Gabriele Sumathy, Vidya Kanellakis, Christoforos Nikolakopoulos, George Robotics and AI Group Department of Computer Science Electrical and Space Engineering Luleå University of Technology Sweden

Autonomous exploration of cluttered environments requires efficient exploration strategies that guarantee safety against potential collisions with unknown random obstacles. This paper presents a novel approach combining a graph neural network-based exploration greedy policy with a safety shield to ensure safe navigation goal selection. The network is trained using reinforcement learning and the proximal policy optimization algorithm to maximize exploration efficiency while reducing the safety shield interventions. However, if the policy selects an infeasible action, the safety shield intervenes to choose the best feasible alternative, ensuring system consistency. Moreover, this paper proposes a reward function that includes a potential field based on the agent’s proximity to unexplored regions and the expected information gain from reaching them. Overall, the approach investigated in this paper merges the benefits of the adaptability of reinforcement learning-driven exploration policies and the guarantee ensured by explicit safety mechanisms. Extensive evaluations in simulated environments demonstrate that the approach enables efficient and safe exploration in cluttered environments. Copyright © 2025, The Authors. All rights reserved.

关键词： Reinforcement learning

Progress Estimation of an Excavation Pit 39

学校读者我要写书评

暂无评论

Progress Estimation of an Excavation Pit

39th International Symposium on Automation and robotics in Construction, ISARC 2022

作者： Vierling, A. Groll, T. Meckel, D. Heim, K. Walter, D. Scheidhauer, L. Körkemeyer, K. Berns, K. Department of Computer Science Robotics Research Lab University of Kaiserslautern Germany Department of Civil Engineering Construction Management University of Kaiserslautern Germany

ISBN: (纸本)9789526952420

This paper presents a method for automated excavation speed and progress estimation. First, a measure for the progress speed of an excavation pit is taken from the literature and evaluated regarding the possibility for automation. For each possible parameter, an automated extraction algorithm is presented. The used system is an autonomous excavator arm of a backhoe loader where the used hardware and software system is described. Experimental evaluation of the presented approach has been done with the autonomous system for a small trench, including multiple digging cycles. The resulting measurements seem to include some systematic errors which could be identified and suitable sanity checks could be implemented, removing the erroneous measurements. The remaining measurements were used to determine the excavation speed of the autonomous excavator arm and compared to the values of experienced and amateur operators. © 2022 International Association on Automation and robotics in Construction.

关键词： Excavation

Curriculum-based Sample Efficient Reinforcement Learning for Robust Stabilization of a Quadrotor

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Suarez, Fausto Mauricio Lagos Saradagi, Akshit Sumathy, Vidya Kotpalliwar, Shruti Nikolakopoulos, George Robotics and AI group in the Department of Computer Science Electrical and Space Engineering Luleå University of Technology Sweden

This article introduces a curriculum learning approach to develop a reinforcement learning-based robust stabilizing controller for a Quadrotor that meets predefined performance criteria. The learning objective is to achieve desired positions from random initial conditions while adhering to both transient and steady-state performance specifications. This objective is challenging for conventional one-stage end-to-end reinforcement learning, due to the strong coupling between position and orientation dynamics, the complexity in designing and tuning the reward function, and poor sample efficiency, which necessitates substantial computational resources and leads to extended convergence times. To address these challenges, this work decomposes the learning objective into a three-stage curriculum that incrementally increases task complexity. The curriculum begins with learning to achieve stable hovering from a fixed initial condition, followed by progressively introducing randomization in initial positions, orientations and velocities. A novel additive reward function is proposed, to incorporate transient and steady-state performance specifications. The results demonstrate that the Proximal Policy Optimization (PPO)based curriculum learning approach, coupled with the proposed reward structure, achieves superior performance compared to a single-stage PPO-trained policy with the same reward function, while significantly reducing computational resource requirements and convergence time. The curriculum-trained policy’s performance and robustness are thoroughly validated under random initial conditions and in the presence of disturbances. © 2025, CC BY.

关键词： Reinforcement learning

REMOTELY CONTROLLABLE DNA MICROFLOW 26

学校读者我要写书评

暂无评论

REMOTELY CONTROLLABLE DNA MICROFLOW

26th International Conference on Miniaturized Systems for Chemistry and Life sciences, MicroTAS 2022

作者： Udono, Hirotake Nomura, Shin-Ichiro M. Takinoue, Masahiro Department of Computer Science School of Computing Tokyo Institute of Technology Japan Department of Robotics Graduate School of Engineering Tohoku University Japan

ISBN: (纸本)9781733419048

Despite an emerging interest in fluid manipulation within confined environment such as biological cells and micro-scale biomimics, remotely controllable micro-scale flow remains unexplored. However, intracellular microfluidics is not compatible with established top-down microfluidic methods because of limited accessibility. Here, we demonstrate photoresponsive DNA fluid, condensate of DNA nanostructures (DNA motifs), as a versatile mechanistic tool for the remote control of microflow. Insertion site of azobenzene, introduced for photocontrollability, determines an extent of dynamic shift in binding enthalpy between DNA motifs, providing the underpinning mechanism for drastic and facile flow-mode switching. © 2022 MicroTAS 2022 - 26th International Conference on Miniaturized Systems for Chemistry and Life sciences. All rights reserved.

关键词： Remote control

Resiliency in Space Autonomy: a Review

学校读者我要写书评

暂无评论

Current robotics Reports 2023年第1期4卷 1-12页

作者： Banerjee, Avijit Mukherjee, Moumita Satpute, Sumeet Nikolakopoulos, George Robotics and AI Department of Computer science Electrical and Space Engineering Luleå University of Technology Luleå Sweden

The article provides an extensive overview on the resilient autonomy advances made across various missions, orbital or deep-space, that captures the current research approaches while investigating the possible future direction of resiliency in space autonomy. In recent years, the need for several automated operations in space applications has been rising, that ranges from the following: spacecraft proximity operations, navigation and some station keeping applications, entry, decent and landing, planetary surface exploration, etc. Also, with the rise of miniaturization concepts in spacecraft, advanced missions with multiple spacecraft platforms introduce more complex behaviours and interactions within the agents, which drives the need for higher levels of autonomy and accommodating collaborative behaviour coupled with robustness to counter unforeseen uncertainties. This collective behaviour is now referred to as resiliency in autonomy. As space missions are getting more and more complex, for example applications where a platform physically interacts with non-cooperative space objects (debris) or planetary bodies coupled with hostile, unpredictable, and extreme environments, there is a rising need for resilient autonomy solutions. Resilience with its key attributes of robustness, redundancy and resourcefulness will lead toward new and enhanced mission paradigms of space missions.

关键词：

A Deep Learning-Based Bengali Visual Question Answering System 25

学校读者我要写书评

暂无评论

A Deep Learning-Based Bengali Visual Question Answering Syst...

25th International Conference on computer and Information Technology, ICCIT 2022

作者： Rafi, Mahamudul Hasan Islam, Shifat Hasan Imtiaz Labib, S.M. Hasan, Sm Sajid Shah, Faisal Muhammad Ahmed, Sifat Ahsanullah University of Science and Technology Department of Computer Science and Engineering Dhaka Bangladesh Bangladesh University of Engineering and Technology Department of Computer Science and Engineering Dhaka Bangladesh Robotics & Artificial Intelligence Environmental Intelligence & Innovation Co. Ltd. Tokyo Japan

ISBN: (纸本)9798350346022

Visual Question Answering (VQA) is a challenging task in Artificial Intelligence (AI), where an AI agent answers questions regarding visual content based on images provided. Therefore, to implement a VQA system, a computer system requires complex reasoning over visual aspects of images and textual parts of the questions to anticipate the correct answer. Although there is a good deal of VQA research in English, Bengali still needs to thoroughly explore this area of artificial intelligence. To address this, we have constructed a Bengali VQA dataset by preparing human-annotated question-answers using a small portion of the images from the VQA v2.0 dataset. To overcome high linguistic priors that hide the importance of precise visual information in visual question answering, we have used real-life scenarios to construct a balanced Bengali VQA dataset. This is the first human-annotated dataset of this kind in Bengali. We have proposed a Top-Down Attention-based approach in this study and conducted several studies to assess our model's performance. © 2022 IEEE.

关键词： Deep learning