Robotics, Computer Vision and Intelligent Systems : First International Conference, ROBOVIS 2020, Virtual Event, November 4-6, 2020, and Second International Conference, ROBOVIS 2021, Virtual Event, October 27-28, 2021, Revised Selected Papers

material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Preface Organization Contents Multi-sensorial Environment Perception in Urban Environment 1 Introduction 2 Lidar Based Urban Scene Analysis 2.1 Ground-Obstacle Classification 2.2 Fast Object Separation and Bounding Box Estimation 2.3 Deep Learning Based Object Recognition in the RMB Lidar Data 2.4 Semantic MLS Point Cloud Classification with a 3D CNN Model 2.5 Multimodal Point Cloud Registration 2.6 Frame Level Cross-modal Change Detection 2.7 Automatic Targetless Lidar-Camera Calibration 3 Evaluation of the Proposed Algorithms 3.1 Experiments on RMB Lidar Based Object Perception 3.2 Evaluation of Semantic MLS Point Cloud Segmentation 3.3 Experiments on Point Cloud Registration 3.4 Experiments on Multimodal Change Detection 3.5 Evaluation of Lidar-Camera Calibration 4 Conclusions of the Paper References An Extended Visual Intelligence Scheme for Disassembly in Automated Recycling Routines 1 Introduction 2 Related Work 3 Methods 3.1 Datasets 3.2 Wire Segmentation 3.3 Screw Classification 4 Bookkeeping 5 Experimental Evaluation 5.1 Evaluation Method 5.2 Generalization 6 Discussion and Conclusion References Multi-agent Path Finding and Acting with Small Reflex-Based Mobile Robots 1 Introduction 1.1 Contribution 2 Overview of MAPF 2.1 Classical Discrete MAPF 2.2 Cumulative Objectives in MAPF 2.3 Real-World Complications 3 Optimal MAPF Solvers 3.1 SAT Based Solver 4 Physical Robot: Ozobot Evo 4.1 Important Hardware 4.2 Movement and Line Following 5 Novel Approach 5.1 ESO-Nav: Reflex-Based Navigation 5.2 Improvements from Previous Work 5.3 Expected Problems 6 From Discrete Planning to Physical Acting 6.1 Overview 6.2 Environment 6.3 MAPF Solver 6.4 Acting via Path Animation 6.5 Modifying Ozobot Behavior 7 Experimental Evaluation 7.1 Maps 7.2 Results 7.3 Compilation-Based Approach 8 Conclusion References Design and Experimental Analysis of an Adaptive Cruise Control 1 Introduction 2 Adaptive Cruise Control 2.1 Fuzzy Logic Approach 2.2 Fuzzy Logic-Based ACC 2.3 Designing of Low-Level Control 3 Experimental Setup 3.1 AutoMiny 3.2 Lab Setup 3.3 Steering Control 4 Evaluation 4.1 Steering Control 4.2 Distance Control Mode 4.3 Speed Control Mode 5 Conclusions References Human-Likeness Indicator for Robot Posture Control and Balance 1 Introduction 1.1 Overview 1.2 Description of the Approach 1.3 Human Data-Set 1.4 Performance Indicator 2 Materials and Methods 2.1 Humanoid Robot 2.2 Control System 2.3 Test Protocol 3 Results 4 Discussion 5 Conclusions References Visual Guidance of an On-Orbit Free-Floating Manipulator Using a Mobile Camera 1 Introduction 2 Robotic Spacecraft 3 Visual Servoing of a Manipulator and Estimation of the Target Motion in the Image Space 3.1 Visual Controller Using the Robotic Camera 3.2 Estimation of the Target Spacecraft Motion in the Image Space 3.3 Target Spacecraft Motion Estimation 4 On-Orbit Robot Manipulator Visual Servoing 5 Results 5.1 Visual Controller to Guide the Manipulator in a Single Movement 5.2 Visual Controller to Guide the Manipulator Following Complex Motion Directions 6 Conclusions References Semantic Segmentation of Marine Species in an Unconstrained Underwater Environment 1 Introduction 2 Related Work 3 Materials and Methods 3.1 The Underwater Fish Observatory 3.2 Camera Setup 3.3 Image Data 3.4 PSPNet 3.5 Intersection-over-Union (IoU) 3.6 Optimization Criteria 4 Experiments 4.1 Setup 4.2 Results 4.3 Discussion 4.4 Conclusion References Social Embodiment of Companion Robots in Smart Spaces: IoRT for Independent Living 1 Introduction 2 Related Work 2.1 Centralized Operation and Integration 2.2 Embodied Intelligence and Social Embodiment 3 Proposed Framework 4 Integrating AR Smart Glass with SCRs 4.1 Establishing Bidirectional Communication Between ROS and AR Smart Glass 4.2 Developing a HoloLens Application in Unity 4.3 Publishing to ROS Topics 4.4 Audio and Image Streaming Through ROS Topics 5 Deep Learning Inference on the Edge 5.1 Analysis of Compute Time 5.2 CNN Architecture for FER 5.3 Miro and Jetson Nano Integration: Deep Learning Inference on the Edge 6 Smart Home Integration with Miro-e 7 Remote and Local Supervision of SCRs Through Microsoft's HoloLens 2 7.1 Centralisation: Controlling an SCR from a Remote Location 7.2 Scaleability: Controlling Multiple SCRs Simultaneously 8 Conclusion References Rapid Structure from Motion Frame Selection for Markerless Monocular SLAM 1 Introduction 1.1 New Contributions 2 Related Work 3 Monocular SLAM Map Initialization 3.1 The Essential Matrix 3.2 The Fundamental Matrix 3.3 The Homography Matrix 3.4 Triangulation 4 Model Configurations 4.1 Data Collection 4.2 Labeling Criteria 5 Results 6 Conclusion References Detecting Tool Keypoints with Synthetic Training Data 1 Introduction 2 Related Work 2.1 Keypoint Detection 2.2 Synthetic Training Data 3 Training Data 3.1 Generating Scenes 3.2 Rendering 4 Method 4.1 Network Architecture 4.2 Training on Synthetic Data 5 Results 5.1 Tool Dataset 5.2 Comparison to Other Methods 5.3 Impact of Generalization Techniques 6 Conclusion References Generating Versatile Training Samples for UAV Trajectory Prediction 1 Introduction and Related Work 2 Prediction Models 3 Synthetic Data Generation 4 Evaluation & Analysis 5 Discussion 6 Conclusion References Author Index This volume constitutes the papers of two workshops which were held in conjunction with the First International Conference on Robotics, Computer Vision and Intelligent Systems, ROBOVIS 2020, Virtual Event, in November 4-6, 2020 and Second International Conference on Robotics, Computer Vision and Intelligent Systems, ROBOVIS 2021, Virtual Event, in October 25-27, 2021. The 11 revised full papers presented in this book were carefully reviewed and selected from 53 submissions