According to the latest reports, although electric car manufacturer Tesla’s driver-assisted driving system Autopilot is being investigated by U.S. regulators for alleged multiple crashes, there are still thousands of car owners testing the company’s latest on public roads.
Version 10.0.1 of the Full Self Driving (FSD) beta software. The latest research from the Massachusetts Institute of Technology (MIT) proves: Although the system is called FSD, it is not actually an automatic driving system, but a more advanced driver-assisted driving system (ADAS), and it may not actually be an automatic driving system. So safe.
Researchers studied saccade data from 290 human subjects when they took over Autopilot and found that drivers may become inattentive when using part of the automated driving system. The study reads: “The driver’s visual behavior patterns will change before and after taking over Autopilot.
Compared with steering to manual driving, the driver usually pays less attention to the road before taking over the control of the vehicle and pays more attention to things that are not related to driving. Field. Before the takeover, the proportion of saccades where the driver’s eyes were off the road was higher, and this cannot be compensated by the longer saccades ahead.
Tesla CEO Elon Musk said that not everyone who paid for the FSD software has access to the FSD beta, even though the software promises to provide more autonomous driving features. First, Tesla will use telemetry data to capture individual driving behavior within seven days to ensure that drivers still maintain sufficient attention. This data can also be used to support a new safety rating page, which tracks the owner’s vehicle and links it to their insurance.
This MIT study provides evidence that drivers may not use Tesla Autopilot as recommended. Since Autopilot includes safety features such as traffic-aware cruise control and automatic steering, drivers’ attention becomes less concentrated and their hands move more away from the steering wheel.
The researchers found that this type of behavior may be the result of a misunderstanding of Autopilot’s functions and limitations. This will be reinforced when Autopilot performs well. With more automated driving operations, drivers will naturally become bored after trying to maintain visual and physical alertness, which can cause further inattention.
The study is titled Natural saccade behavior model before and after Tesla Autopilot takeover, and it tracked Tesla Model S and Model X owners in the Boston area for a year or more. These vehicles are equipped with a real-time intelligent driving environment record data collection system, which continuously collects data from the CAN bus, GPS and three 720P cameras.
These sensors provide information such as vehicle kinematics, driver-vehicle controller interaction, mileage, location, driver posture, face, and field of vision in front of the vehicle. MIT collected nearly 800,000 kilometers of data.
The focus of this research is not to bombard Tesla, but to advocate the development of an attendance management system for drivers, which can provide real-time feedback to drivers or adjust automated functions to suit the driver’s attention level. Currently, Autopilot uses a sensor system to monitor the driver, but it does not monitor the driver’s attention through the eye or head tracking.
The researchers behind this study developed a saccade behavior model based on natural data, which can help understand the characteristics of drivers’ attention shifts under autonomous driving conditions and support the development of solutions to ensure that drivers are fully focused during driving tasks. This can not only help the driver monitoring system deal with atypical gaze changes but also serve as a benchmark for studying the impact of automation on driver behavior safety.
Companies like Seing Machines and Smart Eye have partnered with automakers such as General Motors, Mercedes-Benz, and Ford to introduce camera-based driver monitoring systems for cars equipped with ADAS, but also to solve the problem of drinking and driving. Other problems are caused by dangerous driving. This technology already exists. The question is will Tesla use it?
