It's probably fair to say that off-road driving is more fun than the road bits. When the self-driving car revolution begins, people will mainly use autonomy for the boring bits like commuting and highway miles, and will likely handle the off-road portions themselves.
But one of the hidden consequences of a self-driving city is that people will stop getting driver's licenses. Indeed, they'll stop learning to drive altogether in huge numbers. Within a generation or two, driving will be a hobby as eccentric as horse riding.
So, autonomous cars will need to learn the intricacies of off-roading at some point, and Land Rover has begun the process of developing the technology that'll let four-wheel-drives become four-wheel-self-drives with level 4 and 5 off-road autonomous capabilities.
The key issues will be topography and surface traction. While roads are uniformly flat and generally pretty grippy, off-road driving requires you to manage all four wheels in a dynamic environment. Rocks, ruts, logs, uneven slopes and puddles of indeterminate depth are all obstacles that will need to be dealt with.
Land Rover's autonomous SUV project Cortex aims to address these issues with a multi-sensor approach, using acoustic, video, radar and LiDAR systems to form a combined picture of what the terrain is doing. These systems will have to work in an even wider array of scenarios than the ones in street cars – dirt, mud, big rocks, small rocks, snow, ice and fog all play a much bigger part in off-roading and could foul and blind the sensors in a hundred different ways.
The Cortex team, joined by radar and sensor researchers from the University of Birmningham, as well as machine learning experts from Myrtle AI, are preparing to begin physical testing, sensor optimization and algorithm development at a number of UK off-road tracks.
Source: Jaguar Land Rover
