Tyre-Road Friction Coefficient Estimation using Maximum Entropy
Lead Research Organisation:
Loughborough University
Department Name: Wolfson Sch of Mech, Elec & Manufac Eng
Abstract
Current trends toward automation include the automation of heavy goods vehicles. Heavy goods vehicles have long stopping distances which depend on the state of the road ahead of them. Computer vision can be used to determine the properties of the road ahead of the vehicle, allowing the possibility of determining where the vehicle should go in an emergency stopping manoeuvre.
However, the estimate provided by computer vision of the properties of the road is an estimate, meaning that it has some uncertainty associated with it. Without taking this uncertainty into account, the actual stopping distance of the vehicle might be much longer than anticipated.
The internal architecture for making vehicle motion planning decisions, integrating estimates of friction properties from the computer vision system, has not been clearly defined. In addition, a method is required for defining and understanding the uncertainty of the friction property estimates from the computer vision system and building this into the stopping distance estimation algorithm.
This overseas travel grant seeks to fund a visit to Chalmers University of Technology in Gothenburg, Sweden. The grant will enable close collaboration between Dr Midgley and the vehicle dynamics group at Chalmers. It will also allow technical discussions with Volvo Trucks, who have an interest in the area of heavy vehicle automation. Dr Midgley will be able to work intensively on the architecture and distance estimation problems, while communicating freely with researchers from Chalmers and Volvo Trucks, and build links for future collaborative research projects.
However, the estimate provided by computer vision of the properties of the road is an estimate, meaning that it has some uncertainty associated with it. Without taking this uncertainty into account, the actual stopping distance of the vehicle might be much longer than anticipated.
The internal architecture for making vehicle motion planning decisions, integrating estimates of friction properties from the computer vision system, has not been clearly defined. In addition, a method is required for defining and understanding the uncertainty of the friction property estimates from the computer vision system and building this into the stopping distance estimation algorithm.
This overseas travel grant seeks to fund a visit to Chalmers University of Technology in Gothenburg, Sweden. The grant will enable close collaboration between Dr Midgley and the vehicle dynamics group at Chalmers. It will also allow technical discussions with Volvo Trucks, who have an interest in the area of heavy vehicle automation. Dr Midgley will be able to work intensively on the architecture and distance estimation problems, while communicating freely with researchers from Chalmers and Volvo Trucks, and build links for future collaborative research projects.
Publications
| Description | Heavy Goods Vehicles (HGV) are more massive than standard road vehicles, they also travel across a wider range of different road surfaces and different driving conditions. At the present time there is widespread interest in using datasets to enable autonomous driving, however, these datasets cannot be expanded to HGV due to different viewing angles of onboard vision systems and also due to the broader range of conditions where driving takes place. This travel grant has enabled two visits to Chalmers University and Volvo trucks in Gothenburg, Sweden for Dr Midgley and also Dr Justham, who joined this grant in the latter part of the funding due to Dr Midgley moving institution. Through these technical meetings, novel ideas have been discussed and technical knowledge sharing has taken place. This has facilitated an updated dataset, which is more appropriate for HGV, to be developed and tested. In addition, open sharing of best practise has taken place as well as future collaboration mapping for ongoing work. |
| Exploitation Route | The team of researchers has now expanded to two groups within Chalmers University - both Vehicle Dynamics and Mechatronics - Volvo Trucks and the University of New South Wales, Australia. The team intend to publish work together and develop some collaborate international research initiatives moving forwards. |
| Sectors | Digital/Communication/Information Technologies (including Software),Transport |
| URL | https://arxiv.org/pdf/2303.05927.pdf |
| Description | HGV are part of the backbone of goods distribution globally. However, increasingly autonomous driving and autonomous vehicle technologies are being developed, which are not suitable or generalised for HGV as well as cars. The impact of this work has been the development of HGV relevant datasets, which should help facilitate ongoing development of intelligent driver assistance and the estimation of the friction of any specific road surface. This work has been very well received by Volvo Trucks and they are investigating how we might implement this work on a more general scale for HGV across their business. |
| First Year Of Impact | 2022 |
| Sector | Transport |
| Impact Types | Societal,Economic |
| Description | Chalmers |
| Organisation | Chalmers University of Technology |
| Country | Sweden |
| Sector | Academic/University |
| PI Contribution | We are contributing a novel approach to classifying surfaces and determining their friction coefficient from dash-mounted monocular cameras. |
| Collaborator Contribution | Expertise and engineer time toward building a comprehensive shared endeavour. |
| Impact | A willingness to work together to develop a unified approach toward friction estimation from visual surface information. |
| Start Year | 2021 |