Self-Driving Google Cars Aim to Steer Clear of Bicyclists

Google Inc. reports that part of its autonomous vehicle development program involves ensuring the safety of bicyclists on shared roads.

The company outlines several such initiatives in its latest monthly update on driver-assist technologies. These include establishing spacing guidelines for passing bicyclists and identifying biker hand signals for turning. Advanced algorithms also are being designed to help recognize common riding behaviors and remembering hand signals given well in advance of when a specified turn is initiated, Google says.

If a self-driving vehicle’s sensors detect a parallel-parked car with an open door near a cyclist, the host car is programmed to slow down or move over to give the rider enough room to steer around the obstacle.

Using machine learning technology, Google says its cars can recognize a wide variety of colors, sizes and types of bikes, ranging from unicycles to tandems and so-called conference bikes that accommodate as many as seven riders. The software also can identify and track multiple bikes riding close together or apart during the day or night.

Google is testing 24 specially equipped Lexus RX450h SUVs and 34 prototype vehicles on the roads in Arizona, California and Texas. The company says the fleet averages of about 15,000 miles of autonomous driving a week with a cumulative total of 1.7 million miles to date.