Optimizing the deep learning framework for robotic autonomous navigation (original) (raw)
Related papers
This review article presents recent advancements in deep learning methodologies and applications for autonomous nav- igation. It analyzes state-of-the-art deep learning frameworks used in tasks like signal processing, attitude estimation, ob- stacle detection, scene perception, and path planning. The implementation and testing methodologies of these approaches are critically evaluated, highlighting their strengths, limita- tions, and areas for further development. The review em- phasizes the interdisciplinary nature of autonomous naviga- tion and addresses challenges posed by dynamic and com- plex environments, uncertainty, and obstacles. With a par- ticular focus on mobile robots, self-driving cars, unmanned aerial vehicles, and space vehicles to underscore the impor- tance of navigation in these domains. By synthesizing find- ings from multiple studies, the review aims to be a valuable resource for researchers and practitioners, contributing to the advancement of novel approaches....
DEEP LEARNING APPROACHES FOR AUTONOMOUS VEHICLE NAVIGATION IN URBAN ENVIRONMENTS
IAEME PUBLICATION, 2022
The application of Deep Reinforcement Learning has opened up new avenues for us to explore when it comes to resolving difficult control and navigation-related problems. This article showcases the autonomous navigation and obstacle avoidance capabilities of self-driving autos through the use of Deep Reinforcement Learning. A virtual vehicle operating in a city environment has these features installed after they have been merged with Deep Q Network. This method takes as input data from two types of sensors: one behind the car, which is a video sensor, and another, which is a laser sensor. Furthermore, it builds a model of an affordable high-speed vehicle that can use the same algorithm in real-time, on top of all that. The design integrates a camera and a Hokuyo Lidar sensor in the front of the vehicle. Deep learning algorithms are executed on the basis of sensor inputs by means of an embedded graphics processing unit (NvidiaTX2).
Deep Learning Based Motion Planning For Autonomous Vehicle Using Spatiotemporal LSTM Network
2018 Chinese Automation Congress (CAC), 2018
Motion Planning, as a fundamental technology of automatic navigation for autonomous vehicle, is still an open challenging issue in real-life traffic situation and is mostly applied by the model-based approaches. However, due to the complexity of the traffic situations and the uncertainty of the edge cases, it is hard to devise a general motion planning system for autonomous vehicle. In this paper, we proposed a motion planning model based on deep learning (named as spatiotemporal LSTM network), which is able to generate a real-time reflection based on spatiotemporal information extraction. To be specific, the model based on spatiotemporal LSTM network has three main structure. Firstly, the Convolutional Long-short Term Memory (Conv-LSTM) is used to extract hidden features through sequential image data. Then, the 3D Convolutional Neural Network(3D-CNN) is applied to extract the spatiotemporal information from the multi-frame feature information. Finally, the fully connected neural networks are used to construct a control model for autonomous vehicle steering angle. The experiments demonstrated that the proposed method can generate a robust and accurate visual motion planning results for autonomous vehicle.
Autonomous Navigation in Dynamic Environments: Deep Learning-Based Approach
2021
Mobile robotics is a research area that has witnessed incredible advances for the last decades. Robot navigation is an essential task for mobile robots. Many methods are proposed for allowing robots to navigate within different environments. This thesis studies different deep learningbased approaches, highlighting the advantages and disadvantages of each scheme. In fact, these approaches are promising that some of them can navigate the robot in unknown and dynamic environments. In this thesis, one of the deep learning methods based on convolutional neural network (CNN) is realized by software implementations. There are different preparation studies to complete this thesis such as introduction to Linux, robot operating system (ROS), C++, python, and GAZEBO simulator. Within this work, we modified the drone network (namely, DroNet) approach to be used in an indoor environment by using a ground robot in different cases. Indeed, the DroNet approach suffers from the absence of goal-orien...
End-to-End Deep Learning for Steering Autonomous Vehicles Considering Temporal Dependencies
2017
Steering a car through traffic is a complex task that is difficult to cast into algorithms. Therefore, researchers turn to training artificial neural networks from front-facing camera data stream along with the associated steering angles. Nevertheless, most existing solutions consider only the visual camera frames as input, thus ignoring the temporal relationship between frames. In this work, we propose a Convolutional Long Short-Term Memory Recurrent Neural Network (C-LSTM), that is end-to-end trainable, to learn both visual and dynamic temporal dependencies of driving. Additionally, We introduce posing the steering angle regression problem as classification while imposing a spatial relationship between the output layer neurons. Such method is based on learning a sinusoidal function that encodes steering angles. To train and validate our proposed methods, we used the publicly available Comma.ai dataset. Our solution improved steering root mean square error by 35% over recent method...
Deep learning for self-driving cars
Proceedings of the 1st International Workshop on Software Engineering for AI in Autonomous Systems, 2018
Self-driving cars are a hot research topic in science and technology, which has a great influence on social and economic development. Deep learning is one of the current key areas in the field of artificial intelligence research. It has been widely applied in image processing, natural language understanding, and so on. In recent years, more and more deep learning-based solutions have been presented in the field of self-driving cars and have achieved outstanding results. This paper presents a review of recent research on theories and applications of deep learning for self-driving cars. This survey provides a detailed explanation of the developments of self-driving cars and summarizes the applications of deep learning methods in the field of self-driving cars. Then the main problems in self-driving cars and their solutions based on deep learning methods are analyzed, such as obstacle detection, scene recognition, lane detection, navigation and path planning. In addition, the details of some representative approaches for self-driving cars using deep learning methods are summarized. Finally, the future challenges in the applications of deep learning for self-driving cars are given out.
Image Recognition Based Autonomous Driving: A Deep Learning Approach
Autonomous vehicle (AV) is a broad field in artificial intelligence which has seen monumental growth in the past decade and this had a significant impact in bridging the gap between the capability the intelligence of human and the efficiency of machines. With millions of people losing their lives, or have being a victim of road traffic accidents. There is a need to find a suitable algorithm for a navigation system in an autonomous vehicle with the purpose of help mitigate the traffic rule violation that most human drivers make that lead leads to traffic accidents. With both researchers and enthusiasts developing several algorithms for AVs, this field has been split into several modules which continually broaden the scope of AV's technology. In this paper, we focus on the lane navigation which has an important part of the AV movement on the road. Here lane decision making is optimized by using deep learning techniques in creating a Neural Network model that focuses on generating steering commands by taking an image the road mapped out with lane markings. The navigation aid is a frontfacing camera mounted and images from the camera are used to compute steering commands. The end to end learning scheme was developed by Nvidia cooperation to train a model to compute steering command from a front-facing camera. The model does not focus on detecting the lane but only generating the appropriate command for steering AVs' on the road. This focus on one objective of the model helps in maximizing the potential of better accuracy in lane navigation of our AVs. The modeled car navigates through the designed lanes accurately with the level of intelligence the car shows in maneuvering through the lanes shows this method is more suitable in lane navigation.
Deep Learning for Visual Navigation of Unmanned Ground Vehicles A review
The capabilities that Artificial Intelligence and Computer Vision can provide to intelligent robotic systems is well recognized and as a result it is the subject of topical research in recent years. This paper will provide a broad review of the progress which has been made in applying deep learning and vision sensor data for the autonomous navigation of unmanned ground vehicles (UGVs). The current state-of-the-art techniques are compared in terms of their performance, implementation and deployment and performance. An outline of some of the most popular types of computer vision techniques is provided, as well as insights into how the recent availability of 3D vision systems can be exploited in the domain.
Application of Deep Learning to Autonomous Robotic Car
International Journal of Computer Applications , 2021
Autonomous machines are becoming prevalent, even more so the advent of autonomous vehicles. While autonomous cars have been around for some time, the endless innovations in this domain have led to the removal of human-in-the loop, hence constantly seeking to remove human input while delivering optimal result. However, safety is a major concern, and users are wary of leaving safety level decisions to machines. There is a rise in road accident caused by autonomous cars, while some have blamed it on human's total trust in machines, and researchers have called for the development of human-level accurate algorithms to tackle decision making using state-of-the-art techniques. Therefore, this paper seeks to use computer vision leveraging on deep learning techniques to detect pedestrians, traffic signs, important objects, and lane lines to infer crucial driver decisions.Mask R-Convolutional Neural Network (CNN) was used for object classification with the aid of transfer learning saving the hassle of training and GPU times. A simple method for collecting data was applied using a wide-anglecamera and using Google TPU to perform real time object recognition without the need for a GPU enabled machine.