Aim of the DRIVSCO EU-research project was to investigate feasibility of creating advanced driver assistance systems (ADAS) that would adapt to each driver's particular driving style and improving safety by alerting the driver of possible hazardous situations.
Following is the official description of the project: "Most technical systems, for example cars, must work reliably at key-turn. Therefore, such systems almost always employ conventional control strategies. Biological systems, on the other hand, learn. In the beginning they are functional only at a very basic level from which they improve their skills. No-one would, however, want to use a learning car, which could in the beginning barely steer. Thus, learning techniques have not really entered turn-key applications so far.
The goal of DRIVSCO is to devise, test and implement a strategy of how to combine adaptive learning mechanisms with conventional control, starting with a fully operational human-machine interfaced control system and arriving at a strongly improved, largely autonomous system after learning, that will act in a proactive way using different predictive mechanisms.".
System Schematics
Following figure shows how the DRIVSCO system integrates in the vehicle infrastructure using CAN-bus. As it can be understood, integrating data from on-board sensors with visual cues allows for generation of more robust and coherent descriptors that can be used by the ADAS system.
In order to assess performance of the system and to verify how well the system can detect hazardous situations, several aspects of the ADAS were accessible through a GUI. The figure below shows data for steering angle history/prediction.
Project Details
- Beginning date: 01 Feb 2006
- Ending date: 31 Jul 2009
- Funding: European Commission (FP6-IST-FET, contract 016276-2)
My Role
My role in the project was related to researching and implementing robust stereo disparity and optical flow calculation methods that in later projects could be used, amongst other things, for time-to-impact and related approximations. During the project I published several scientific papers:
- A Method for Sparse Disparity Densification using Voting Mask Propagation
- Disparity Disambiguation by Fusion of Signal- and Symbolic-level Information
- Increasing Efficiency in Disparity Calculation
During this time other ideas were seeded that later on were published.
Results
In the following there are two videos demonstrating some of the results from my research related to DRIVSCO. The first video shows a video of stereo disparity (inversely proportional to distance of objects from the camera) shot using the cameras installed in the DRIVSCO test vehicle. The darker the colour the closer the object is to the camera.
High quality version of the video in MP4 format is available from here
The second video shows a video of optical flow calculated for the same scene. Direction of movement is colour codified.
High quality version of the video in MP4 format is available from here
