Media | rFpro

Training, testing, validating Deep Learning Autonomous Driving

Creating digital road models of public test routes from kinetic LiDAR, followed by a live demonstration of the main features in rFpro for the testing, training and testing of Deep Learning Autonomous Driving:

  1. real road models, not synthetic data – avoiding the patterns inherent in synthetic data that damage DNN training
  2. correlation with HD maps – map the road signs detected by your perception algorithms to HD map ground truth
  3. detail road surface for radar sensor models – localisation from road signature correlation
  4. Add traffic – swarm and programmed traffic
  5. Add pedestrians – swarm and programmed pedestrians, cyclists, animals etc
  6. Run real-time and add human test drivers – mix stochastic, unpredictable, error-prone human behaviour live.
  7. Massively parallel, multiple ego cars, multiple sensor feeds, multiple human drivers.


This is Jaguar Land Rover’s Test Route, with live traffic, showing an example of a Digital Road Model of a public road test route.  Due to its high quality it is possible to achieve correlation not just for vehicle dynamics testing, but also for ADAS & Autonomous testing that exercises the toolchain from sensor model through to ground truth validation and ultimately a driver in the loop.


This video compresses an entire day, from dawn to dusk, with four changes of weather into just 10 seconds to illustrate the variety of conditions that may be simulated when using a driving simulator with rFpro for testing and validation, particularly of ADAS and Autonomous systems.

A quick introduction to rFpro showing highlights of the Nordschleife, over 4 man-years effort were required to reproduce the circuit.  The entire 20.8km circuit’s road surface has been built with 1cm resolution in x and y accurate to less than 1 millimetre.

rFpro Dawn To Dusk – using rFpro to test ADAS sensors in different lighting conditions. Shadows and direct sunlight can make the white lines, road signs and other features difficult to detect. rFpro allows you to exercise your entire ADAS toolchain, from sensor through to human driver in a wide variety of conditions.

rFpro is used by Automotive OEMs and Motorsport manufacturers to assist with the development of chassis, powertrain, driveline and safety systems.

rFpro is being used by most of the F1 teams and NASCAR manufacturers to assist with development and testing of new cars and setups. This video shows an example of an F1 circuit