In core computer vision tasks, we have witnessed significant advances in object detection, localisation and tracking. However, there are currently no methods to detect, localize and track objects in road environments, and taking into account real-time constraints. In this paper, our objective is to develop a deep learning multi object detection and tracking technique applied to road smart mobility. Firstly, we propose an effective detector-based on YOLOv3 which we adapt to our context. Subsequently, to localize successfully the detected objects, we put forward an adaptive method aiming to extract 3D information, i.e., depth maps. To do so, a comparative study is carried out taking into account two approaches: Monodepth2 for monocular vision and MADNEt for stereoscopic vision. These approaches are then evaluated over datasets containing depth information in order to discern the best solution that performs better in real-time conditions. Object tracking is necessary in order to mitigate the risks of collisions. Unlike traditional tracking approaches which require target initialization beforehand, our approach consists of using information from object detection and distance estimation to initialize targets and to track them later. Expressly, we propose here to improve SORT approach for 3D object tracking. We introduce an extended Kalman filter to better estimate the position of objects. Extensive experiments carried out on KITTI dataset prove that our proposal outperforms state-of-the-art approches.

This paper deals with the development of an Advanced Driver Assistance System (ADAS) for a smart electric wheelchair in order to improve the autonomy of disabled people. Our use case, built from a formal clinical study, is based on the detection, depth estimation, localization and tracking of objects in wheelchair’s indoor environment, namely: door and door handles. The aim of this work is to provide a perception layer to the wheelchair, enabling this way the detection of these keypoints in its immediate surrounding, and constructing of a short lifespan semantic map. Firstly, we present an adaptation of the YOLOv3 object detection algorithm to our use case. Then, we present our depth estimation approach using an Intel RealSense camera. Finally, as a third and last step of our approach, we present our 3D object tracking approach based on the SORT algorithm. In order to validate all the developments, we have carried out different experiments in a controlled indoor environment. Detection, distance estimation and object tracking are experimented using our own dataset, which includes doors and door handles.

For smart mobility, and autonomous vehicles (AV), it is necessary to have a very precise perception of the environment to guarantee reliable decision-making, and to be able to extend the results obtained for the road sector to other areas such as rail. To this end, we introduce a new single-stage monocular real-time 3D object detection convolutional neural network (CNN) based on YOLOv5, dedicated to smart mobility applications for both road and rail environments. To perform the 3D parameter regression, we replace YOLOv5’s anchor boxes with our hybrid anchor boxes. Our method is available in different model sizes such as YOLOv5: small, medium, and large. The new model that we propose is optimized for real-time embedded constraints (lightweight, speed, and accuracy) that takes advantage of the improvement brought by split attention (SA) convolutions called small split attention model (Small-SA). To validate our CNN model, we also introduce a new virtual dataset for both road and rail environments by leveraging the video game Grand Theft Auto V (GTAV). We provide extensive results of our different models on both KITTI and our own GTAV datasets. Through our results, we show that our method is the fastest available 3D object detection with accuracy results close to state-of-the-art methods on the KITTI road dataset. We further demonstrate that the pre-training process on our GTAV virtual dataset improves the accuracy on real datasets such as KITTI, thus allowing our method to obtain an even greater accuracy than state-of-the-art approaches with 16.16% 3D average precision on hard car detection with inference time of 11.1 ms/image on an RTX 3080 GPU.

For smart mobility, autonomous vehicles, and advanced driver-assistance systems (ADASs), perception of the environment is an important task in scene analysis and understanding. Better perception of the environment allows for enhanced decision making, which, in turn, enables very high-precision actions. To this end, we introduce in this work a new real-time deep learning approach for 3D multi-object detection for smart mobility not only on roads, but also on railways. To obtain the 3D bounding boxes of the objects, we modified a proven real-time 2D detector, YOLOv3, to predict 3D object localization, object dimensions, and object orientation. Our method has been evaluated on KITTI’s road dataset as well as on our own hybrid virtual road/rail dataset acquired from the video game Grand Theft Auto (GTA) V. The evaluation of our method on these two datasets shows good accuracy, but more importantly that it can be used in real-time conditions, in road and rail traffic environments. Through our experimental results, we also show the importance of the accuracy of prediction of the regions of interest (RoIs) used in the estimation of 3D bounding box parameters.

On-road behavior analysis is a crucial and challenging problem in the autonomous driving vision-based area. Several endeavors have been proposed to deal with different related tasks and it has gained wide attention recently. Much of the excitement about on-road behavior understanding has been the labor of advancement witnessed in the fields of computer vision, machine, and deep learning. Remarkable achievements have been made in the Road Behavior Understanding area over the last years. This paper reviews 100+ papers of on-road behavior analysis related work in the light of the milestones achieved, spanning over the last 2 decades. This review paper provides the first attempt to draw smart mobility researchers’ attention to the road behavior understanding field and its potential impact on road safety to the whole road agents such as: drivers, pedestrians, stuffs, etc. To push for an holistic understanding, we investigate the complementary relationships between different elementary tasks that we define as the main components of road behavior understanding to achieve a comprehensive understanding of approaches and techniques. For this, five related topics have been covered in this review, including situational awareness, driver-road interaction, road scene understanding, trajectories forecast, driving activities, and status analysis. This paper also reviews the contribution of deep learning approaches and makes an in-depth analysis of recent benchmarks as well, with a specific taxonomy that can help stakeholders in selecting their best-fit architecture. We also finally provide a comprehensive discussion leading us to identify novel research directions some of which have been implemented and validated in our current smart mobility research work. This paper presents the first survey of road behavior understanding-related work without overlap with existing reviews.

The real-time segmentation of sidewalk environments is critical to achieving autonomous navigation for robotic wheelchairs in urban territories. A robust and real-time video semantic segmentation offers an apt solution for advanced visual perception in such complex domains. The key to this proposition is to have a method with lightweight flow estimations and reliable feature extractions. We address this by selecting an approach based on recent trends in video segmentation. Although these approaches demonstrate efficient and cost-effective segmentation performance in cross-domain implementations, they require additional procedures to put their striking characteristics into practical use. We use our method for developing a visual perception technique to perform in urban sidewalk environments for the robotic wheelchair. We generate a collection of synthetic scenes in a blending target distribution to train and validate our approach. Experimental results show that our method improves prediction accuracy on our benchmark with tolerable loss of speed and without additional overhead. Overall, our technique serves as a reference to transfer and develop perception algorithms for any cross-domain visual perception applications with less downtime.