With a string of colorful new Tesla factory openings and already bored investor presentations, the ongoing investigation into the phantom emergency braking incident on Tesla vehicles has been ignored. We want to consider in this article the possible causes of phantom braking and their consequences for the company.
Conventionally, any causes that could cause phantom braking of Tesla cars can be divided into two groups: one-time and mass. By one-time reasons, we mean those reasons that affect some individual components and spare parts of cars installed on a certain series. Mass group means some kind of defect that will affect all or a significant part of Tesla vehicles.
Let’s take a closer look at the possible causes of phantom braking in Tesla cars and their consequences for the automaker and its customers. Let’s start with the implications for the company’s customers. A one-time technical malfunction will entail the recall of cars and the need to replace some structural elements. This is somewhat unpleasant, as it will require time for car owners to replace it, but given the loyalty to the brand, this minor annoyance will go almost unnoticed.
The situation with mass replacement is a little more complicated, which can affect all car owners or some significant part of it. But such a replacement is possible only if the malfunction that causes emergency spontaneous braking of Tesla cars is detected in some central element of the car or in the chosen technology, for example, in autopilot technology.
If a one-time defect is detected in some element of the vehicle’s suspension or braking system, all that remains is to find the cars affected by this incident and replace the required part. In the event of a mass technological defect or a previously unaccounted-for parameter, the replacement may affect some significant parts of the vehicles and may cost the company significant financial resources. And the anti-advertising received as a result of such a replacement in the conditions of falling incomes of the population and the saturation of customers loyal to the company with cars is not at all desirable. Therefore, the early identification of the problem and its elimination is in the interests of everyone, both car owners and Tesla investors.
Given my experience in video surveillance systems, 3D scanners, and software development, I will allow myself to express one of the theories of what exactly could cause the phantom braking of Tesla cars and how to solve this problem if this theory turns out to be correct. A possible cause could be… flies! Yes, you heard right. Flies and other flying insects. It is unlikely that any of us can assume this on our own; experience is needed for such an assumption. For example, our company has such an experience when, while developing software for detecting drones in a video stream, that is, from a video from a camera, we were faced with the fact that a million flies fly in front of a video camera on one drone.
The background of the possible influence of flies on the operation of Tesla’s autopilot system is as follows. A lidar was installed for orientation on the first series of Tesla cars in the surrounding space. This is a device that, like a 3D scanner, scans the surrounding space and creates a three-dimensional picture of space. The resolution of the lidar does not allow it to see small objects, such as flying insects.
However, in later series of vehicles, Tesla began to use a solution based on video cameras in the autopilot system. The system of video cameras installed in Tesla cars has a number of undeniable advantages. And how they achieved such a quality of recognition, which demonstrates the work of the autopilot today, I can’t even imagine, despite my experience in software development. But what about the flies?
Regardless of the quality of the software of the autopilot recognition system, this system has a “bottleneck” in the form of a limitation of any video camera that receives a two-dimensional image of the surrounding world. And one camera cannot determine the distance to the object or its size. In the case of using several video cameras, these tasks can already be solved conditionally, but now this is not important for the issue we are considering. What happens to the video camera if a fly flies into it while the car is moving? The video camera will see an unexpected obstacle and instruct the autopilot system to brake.
Of course, according to the theory of probability, insects that collide with a car while driving will only fall into the lens of a video camera in a small percentage of cases. But this is quite consistent with the statistics of rare cases of phantom braking. So, this theory is quite right to life. How can this issue be resolved? After all, it will have to be solved, because although such cases are rare, this endangers the lives of car owners and other road users.
And levels the pleasant impressions of the owners from using the autopilot. There are several options for solving this problem. This is a return to lidar, which will not respond to insects. Or updating the software, taking into account the correction for flies. This is the technology we have already developed, it is described on the Pterygota Marker website, which allows us to filter insects in the video stream.
We wish Tesla further successful development and the creation of even more incredible solutions, from which you just want to exclaim, damn it, how did they do it?
Pterygota Marker is a device for detecting drones in video streams and machine vision systems, which allows to distinguish between flying insects and moving objects.
