Keeping his promise of revealing major improvements coming soon to the Tesla Autopilot, CEO Elon Musk talked some significant changes that will be making their way to Tesla’s auto driving systems. In a press call that took place a few hours ago, the major topics on the table included radar and fleet learning. Musk also said that thanks to this combination, there should be an almost ‘three fold increase in safety’.
Elaborating upon his plans for improving the safety features of the autopilot system, Musk said that the latest 8.0 Autopilot software update will turn the existing radar system inside Tesla cars into a primary source of information. The information derived from this source will then be deployed towards improving the vehicle’s self-driving features. This change in stand is pretty important, because the Radar was inducted into Tesla vehicles to be merely an aid to the primary camera.
However, the company now appears to be ready to delegate more important jobs to it. As per a Tesla official press release,
While there are dozens of small refinements with Version 8 of our software, described in addendum below, the most significant upgrade to Autopilot will be the use of more advanced signal processing to create a picture of the world using the on-board radar.
The radar was added to all Tesla vehicles in October 2014 as part of the Autopilot hardware suite, but was only meant to be a supplementary sensor to the primary camera and image processing system.
Radars are great. So why did Tesla cars preferred to use the combination of a camera and an image processing system over it? Well, the answer actually lies in the way the radar works and the huge differences that lie between how we perceive the world, and how it looks in radar.
While the emissions from radar can easily travel through fog, dust, rain and snow — which may be one of the reasons it was inducted as a supplementary camera — anything metallic looks and acts like a mirror. Similarly, radars can be used to acknowledge the presence of people, however, they appear as partially translucent. And wood, plastic and other light material are practically transparent, or not-there for all practical purposes, to the radar. That means, that a car operating solely on the would likely refuse to acknowledge the presence of say, a wooden wall, and slam into it — not a pleasant prospect for the occupants of the car.
However, the 8.0 update is taking care of these issues by enhancing the capabilities of the Tesla car hardware. As per the post,
Software 8.0 unlocks access to six times as many radar objects with the same hardware with a lot more information per object.
So basically, the radars would be collecting a lot more data after the update than they were doing before it.
Next up, these snapshots would be assembled together into a cohesive picture. While a single frame won’t be able inform you about whether an object is moving or stationary, a combination of these frames is much more informative and can prevent your car from running into objects.
Finally, while you may easily distinguish between a rise in the road, to the radar, that is just an obstacle. So, the car may just slow down or even stop altogether in order to prevent, what appears to it as a collision. Or conversely, it may just keep on going full speed ahead and bump your head against the roof. The navigation data and height accuracy of the GPS are simply not enough to prevent this.
Tesla has used the same principle that brought the human race to the point where we began making cars that run on electricity — from when we were rubbing rocks to get sparks — and that is, cumulative learning. Initially, the cars wouldn’t take any action when encountering road signs, bridges and other stationary objects, except that they would be mapping the world according to radar. The objects would be noted and uploaded to the Tesla database. With time though, cars would begin making use of this database to improve their driving.
If several cars drive safely past a given radar object, whether Autopilot is turned on or off, then that object is added to the geocoded whitelist.
Also, even if you are trailing the invisible car from Megamind or if there is simply too much fog and the car in front is invisible to radar — by being made of wood, somehow — Tesla has you covered. You car will be able to bounce the radar signal under the vehicle in front – using the radar pulse signature and photon time of flight to distinguish the signal — and make sure that it brakes at the right time.
Speaking on the additional safety provided by these features, Musk said,
I would would imagine that the 8.0 set of improvements, radar, inclusive of the others, probably cuts the accident rates more than in half — that’s my guess. I think it would make the Model S and X by far the safest on the road.
Finally, here are some other less important changes being brought to the workings of its cars by Tesla.
- TACC braking max ramp rate increased and latency reduced by a factor of five
- Now controls for two cars ahead using radar echo, improving cut-out response and reaction time to otherwise-invisible heavy braking events
- Will take highway exit if indicator on (8.0) or if nav system active (8.1). Available in the United States initially
- Car offsets in lane when overtaking a slower vehicle driving close to its lane edge
- Interface alerts are much more prominent, including flashing white border on instrument panel
- Improved cut-in detection using blinker on vehicle ahead
- Reduced likelihood of overtaking in right lane in Europe
- Improved auto lane change availability
- Car will not allow reengagement of Autosteer until parked if user ignores repeated warnings
- Automatic braking will now amplify user braking in emergencies
- In manual mode, alerts driver if about to leave the road and no torque on steering wheel has been detected since Autosteer was deactivated
- With further data gathering, car will activate Autosteer to avoid collision when probability ~100%
- Curve speed adaptation now uses fleet-learned roadway curvature
- Approximately 200 small enhancements that aren’t worth a bullet point
These updates will be deployed via over-the-air basis to Tesla Model S and Model X cars, released after 2013. That means that users won’t have to take their cars anywhere and can simply leave it parked for some time, while Tesla updates its auto-pilot software. Tesla will be rolling the Autopilot 8.0 update to global users within the next couple of weeks.
Meanwhile, the 8.0 update actually appears to be one long step along the path to sustainable autopilot. While Tesla has always stressed that the autopilot feature on it’s cars is not meant for urban areas just yet and instead is more suitable for freeways, the induction of radar as a primary source of information along with the image processing system already there means a significant ramping up of Model S and Model X capabilities.
As Musk noted, perfect safety is impossible. Heck, humans probably drive the vast majority of cars in the world and god knows the trouble we get into, after using our brains and all the other fantastic capabilities we have. So, yeah, Tesla cars may not be the safest thing to go cruising in on a busy lane at the moment. However, the company is working hard and with features like Fleet learning, we are only likely to see their capabilities going upwards.