Abstract
When driving, people make decisions based on current traffic as well as their desired route. They have a mental map of known routes and are often able to navigate without needing directions. Current published self-driving models improve their performances when using additional GPS information. Here we aim to push forward self-driving research and perform route planning even in the complete absence of GPS at inference time. Our system learns to predict in real-time vehicle’s current location and future trajectory, on a known map, given only the raw video stream and the final destination. Trajectories consist of instant steering commands that depend on present traffic, as well as longer-term navigation decisions towards a specific destination. Along with our novel proposed approach to localization and navigation from visual data, we also introduce a novel large dataset in an urban environment, which consists of video and GPS streams collected with a smartphone while driving. The GPS is automatically processed to obtain supervision labels and to create an analytical representation of the traversed map. In tests, our solution outperforms published state of the art methods on visual localization and steering and provides reliable navigation assistance between any two known locations. We also show that our system can adapt to short and long-term changes in weather conditions or the structure of the urban environment. We make the entire dataset and the code publicly available.
Highlights
Nowadays, self-driving cars and intelligent driver-assistant systems heavily relying on vision are emerging in our everyday reality
We introduce the Urban European Driving Dataset (UED), which we make available along with our code and user-friendly application, which we developed for both data collection and real-time usage
We compared to the recent work of Marcu et al, which is the first to formulate the localization task as segmentation and various versions of the current system (LOVis-2DOF, localization from vision system (LOVis), LOVis-reg and LOVis-F), as follows: LOVis-2DOF is the basic localization as segmentation system presented here, without the orientation predicted as the half-circle segmentation; LOVis is the full deep net architecture with both localization and orientation prediction; LOVis-reg has almost the same deep net architecture as
Summary
Self-driving cars and intelligent driver-assistant systems heavily relying on vision are emerging in our everyday reality. Even though we acknowledge that for self-driving cars, additional control modules are necessary (such as the ones based on GPS and LiDAR), it is essential to increase the performances and the perception of the vision-based solutions. Since vision, at both short and long-range distances, is almost always available at relatively low-cost. There are models [1,2] that use visual information to extract highlevel semantics of the traffic scene and decide the steering action conditioned on these representations. On the other hand, collecting data and training is more efficient in end-to-end solutions, which can learn more relevant features
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
