Developments with Tesla’s autopilot system are now enabling their driverless cars to 'see' bikes that are filtering between moving traffic. This is a huge step forward for autonomous vehicles as motorcycles present numerous problems for them.
Not only are bikes small but they tend to move erratically, unlike cars that are large and (with some obvious exceptions) tend to stick to a single lane.
MORE FROM MCN
This makes tracking bikes, and therefore predicting their next moves, very difficult and presents a huge challenge for autonomous vehicle adoption. To combat this problem, Tesla have been working on a more advanced 'neural net' that can identify the presence of numerous vehicles, both large and small, around the car.
In a video of a Tesla Model 3 running the firm's latest Version 9 software, the car can be seen sensing bikes as they filter between vehicles.
Although it's far from perfect, the Tesla sometimes mistakes the bike for a car or places it in the adjacent lane, it certainly shows promise. In cases where the cameras have missed the bikes, the ultrasonic sensors still pick them up, which would result in an audible warning if the driver had put an indicator on or attempted to change lanes.
Considering how difficult some drivers find spotting bikes, the fact a computer manages it is incredibly impressive. However, this does represent one slightly more worrying step towards Judgement Day...