Loading...
Please wait, while we are loading the content...
Similar Documents
Perception och lokalisering för ett Apollobaserat autonomt fordon
| Content Provider | Semantic Scholar |
|---|---|
| Author | Hiller, Julius Hultgren, Daniel Karlsson, Filip Lundberg, Filip Raudberget, Emil Ulmestrand, Jonathan |
| Copyright Year | 2018 |
| Abstract | An imperative part of creating an autonomous vehicle is to localize it and to make it perceive its surroundings in order to navigate in an optimized and secure manner. In this project a localization and perception system has been implemented using a lidar, camera, GNSS-receiver and an inertial measurement unit for use in an autonomous vehicle. Since this usually requires advanced algorithms, the software platform Apollo and an Extended Kalman Filter has been used to process the data. The platform and filter processes data from the sensors to classify objects in the environment and to determine the position and movement of the vehicle. The system managed to identify objects on the road near the vehicle, but under certain circumstances the objects detected did not correlate with the actual object. For example, a traffic cone or a bush was identified as a pedestrian. More testing is required to determine the accuracy of the system. Positioning was often accurate but unstable and sensitive of disturbances in satellite and radio coverage. The accuracy of the orientation of the vehicle was deemed insufficient. This was due to the fact that no sensor for meassuring the absoulte heading was used. This project shows that the advanced technology to make a vehicle autonomous is not only available to the largest organizations anymore, but is also publicly available. The fact that the technology is so accessible may further accelerate the maturity of the technology. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://publications.lib.chalmers.se/records/fulltext/255852/255852.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |