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

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