Google has created a bunch of autonomous self driving cars which are now roaming across the streets. They are doing a good job so far, but just like every other new software, there are plenty of bugs to be fixed.
A hilarious issue was identified a couple of days ago when a professional cyclist had an encounter with one such self-driving car from Google. These cars usually have passengers inside who are responsible for taking notes and submitting a comprehensive report at the end of the journey. However, they don’t have a direct control over the car and the practical issues it faces on the road.
During this particular encounter, a cyclist and the autonomous car met in a four way intersection. The vehicle is equipped with a motion sensor and other sensors that can scan objects in front of it. In order to allow the car to pass, the cyclist did a track stand and waited for the vehicle to take the turn.
A track stand is a position where a rider balances himself on top of a bicycle without placing his legs on the road. It is easy to do it on a gear bicycle because it allows them to keep the pedal in a fixed position for a long time. Being a professional rider, the cyclist did a track stand with ease, but according to Google’s self-driving car, he was about to ride.
According to the biker who shared his experience on a cyclists’ forum, the car came to a halt and expected him to make the first move assuming that he’s in a position to pedal. When he didn’t move, the car started moving and by instinct the person did the same. Both of them had to halt multiple times in order to avoid a collision after which the cyclist took a turn and changed course so as to allow the autonomous car to pass.
In the forum, he added that the people who were seated inside the car took notes and probably elaborated the situation so that the Google car’s programming is properly modified to identify a track stand from a person in motion who’s actually pedaling on a bicycle.
So far, Google’s self-driving cars powered by their Maps service are enjoying a good run. The cars have met with fewer accidents and in most cases, it was because of an error committed by a human rather than the AI. Analysts predict that within 5 to 10 years, these autonomous cars could become ubiquitous.