Project Overview.
VeriCAV has developed an integrated platform to allow Automated Driving Systems to be tested in simulation. As automated vehicle technology becomes increasingly sophisticated, new methods are needed to ensure safety. Simulation offers efficient evaluation of complex, infrequent and unsafe situations.
VeriCAV improves test efficiency when evaluating countless driving scenarios whilst simultaneously incorporating intelligent behaviour into simulated road-users.
Generate Generate
This graph shows the Automated Vehicle navigating the generated test scenario.
Test Generator.
Please select from the options below to generate a test scenario.
Our Objective.
Simulation to support Automated Vehicle deployment.
Testing in simulation is a vital part of a wider set of Automated Vehicle verification activities - offering greater repeatability, efficiency and controllability. Testing in simulation contributes to the overall body of evidence to build the safety case for deployment.
Time Elapsed: Seconds
Selected a gap ahead to merging vehicle.
Complex test space
A simplified selection of variables is used for this demonstration. Other variables that affect these scenarios include relative speeds of the vehicles involved, numbers of driving lanes, driving behaviour of the other roads users, plus many others.
Loading…
The Test generator is now setting up the selected test scenario
The video above demonstrates our simulated test scenario.
Test Scenario.
Test Comparison.
This graph shows the Automated Vehicle navigating the generated test scenario.
Smart Actors.
Realistic Behaviour.
A realistic test of Automated vehicles needs realistic behaviour from the other road users. VeriCAV delivers this via ‘smart actors’ – intelligently moving pedestrians and vehicles that have their own awareness and make their own decisions.
- Pedestrians and Vehicles in the real-world move with a goal and purpose but each displays subtleties - variation in movement as a result of each individual's decision making.
- University of Leeds is using highly innovative approaches including incorporating machine learning techniques to bring this rich behaviour into the test scenarios.
- The smart actor behaviour will be influenced by the scenario and the movement of the Automated Vehicle.
Control driving simulator
Hold left mouse button to accelerate. Right mouse button to decelerate.
Driving Sim.
- Explore and visualise the world of Smart actors.
- This Driving Simulator allows user to control speed of a vehicle as it encounters traffic vehicles turning right across its path.
- Developed by University of Leeds as part of VeriCAV.
- Driving simulator will include increasingly complex Smart Actor behaviour.
- Arrows above the actor vehicles show they are turning. Egg timer indicates they are waiting for space to turn.
Summary & Results.
- Test oracle performs an automated assessment of the Automated Vehicle under test against a range of metrics, including overall safety, good road etiquette, and smooth driving
- For this simple scenario the Automated vehicle has been judged to have made reasonable progress, in a safe manner without inconveniencing other road traffic
- Search space optimisation can now alter the setup of the test scenario to best evaluate the performance of the Automated Vehicle under test.
Plot Showing Test Oracle output for selected Test Scenario
Alternative Scenarios.
Compare Scene Types.
Shows how the other test generator variables effect and impact the outcome of the scenarios.
Selecting the drop down above can show more scenario options.
Interested in Collaboration?
Want to know more?
If you would like to know more about VeriCAV innovation or get in contact with partners click here.
Read More