Deep Learning

Disentangling Object Motion and Occlusion for Unsupervised Multi-frame Monocular Depth

Abstract: Conventional self-supervised monocular depth prediction methods are based on a static environment assumption, which leads to accuracy degradation in dynamic scenes due to the mismatch and occlusion problems introduced by object motions.

Advancing Self-Supervised Monocular Depth Learning with Sparse LiDAR

Abstract: Self-supervised monocular depth prediction provides a cost-effective solution to obtain the 3D location of each pixel. However, the existing approaches usually lead to unsatisfactory accuracy, which is critical for autonomous robots.

Multimodal Semi-Supervised Learning for 3D Objects

Abstract: In recent years, semi-supervised learning has been widely explored and shows excellent data efficiency for 2D data. There is an emerging need to improve data efficiency for 3D tasks due to the scarcity of labeled 3D data.

PSE-Match: A Viewpoint-Free Place Recognition Method With Parallel Semantic Embedding

Abstract: Accurate localization on the autonomous driving cars is essential for autonomy and driving safety, especially for complex urban streets and search-and-rescue subterranean environments where high-accurate GPS is not available. However current odometry estimation may introduce the drifting problems in long-term navigation without robust global localization.

Multi-Scale Fusion With Matching Attention Model: A Novel Decoding Network Cooperated With NAS for Real-Time Semantic Segmentation

Abstract: This paper proposes a real-time multi-scale semantic segmentation network (MsNet). MsNet is a combination of our novel multi-scale fusion with matching attention model (MFMA) as the decoding network and the network searched by asymptotic neural architecture search (ANAS) or MobileNetV3 as the encoding network.

Semantic Digital Surface Map Towards Collaborative Off-Road Vehicle Autonomy

Abstract: The fundamental aspect of unmanned ground vehicle (UGV) navigation, especially over off-road environments, are representations of terrain describing geometry, types, and traversability. One of the typical representations of the environment is digital surface models (DSMs) which efficiently encode geometric information.

Concrete Defects Inspection and 3D Mapping Using CityFlyer Quadrotor Robot

Abstract: The concrete aging problem has gained more attention in recent years as more bridges and tunnels in the United States lack proper maintenance. Though the Federal Highway Administration requires these public concrete structures to be inspected regularly, on-site manual inspection by human operators is time-consuming and labor-intensive.

Driver Drowsiness Behavior Detection and Analysis Using Vision-Based Multimodal Features for Driving Safety

Abstract: Driving inattention caused by drowsiness has been a significant reason for vehicle crash accidents, and there is a critical need to augment driving safety by monitoring driver drowsiness behaviors. For real-time drowsy driving awareness, we propose a vision-based driver drowsiness monitoring system (DDMS) for driver drowsiness behavior recognition and analysis.

Deep Neural Network based Visual Inspection with 3D Metric Measurement of Concrete Defects using Wall-climbing Robot

Abstract: This paper presents a novel metric inspection robot system using a deep neural network to detect and measure surface flaws (i.e., crack and spalling) on concrete structures performed by a wall-climbing robot.

Semantic Metric 3D Reconstruction for Concrete Inspection

Abstract: In this paper, we exploit the concrete surface flaw inspection through the fusion of visual positioning and semantic segmentation approach. The fused inspection result is represented by a 3D metric map with a spatial area, width, and depth information, which shows the advantage over general inspection in image space without metric info.