MIT researchers along with the company Quantum Signal have been testing a new system, which helps cars serve as automotive co-pilots. It is still a while before cars drive by themselves, but they may start off by helping humans become better drivers. MIT has developed a system enabling cars to see traffic cones and other obstacles that a driver may fail to notice, and override instructions to jump red lights. This system allows cars to share driving control and provides a safety blanket.
The system is equipped with a camera and laser rangefinder to spot obstacles. An algorithm identifies safe zones and computes the location of obstacles. MIT believes that the system will be easier and cheaper to implement, as it is simpler than Google’s driverless car fleet that is based on a set of programs and sensors.
These developments are part of an ongoing process by research groups and automakers to bring better intelligence into the vehicle cabin. These systems of different levels of complexity have seen varied results. General Motors is currently building a system that detect other cars in crowded highways and auto applies the brakes. Volkswagen is coming up with a temporary autopilot system that takes over the control of the car in certain circumstances. Europe’s road-train program allows cars to follow each other by connecting them wirelessly.
These systems perform specific tasks based on preset instructions. The MIT system attempts to go beyond this. Instead of programmed rules or robotic instructions, this response system seeks to make a car operate like a human. The system has been test run 1200 times by MIT research scientist Karl Lagnemma and student Sterling Anderson. It has performed well for most part, with a few collisions on account of the quirky hardware configuration.
As Anderson puts it, the system works in line with human thought processes and takes decisions based on certain factors. A human driver on a lane typically identifies the region he/she can use and a safe zone within this region. The driver tries to avoid collisions and stay safe, without bothering to stick to a specific line of traffic. The system divides the entire traffic area into triangles which define safe and unsafe areas. The driver has the control of the car in safe zones and makes the driving choices. The system takes over control where the triangles identify obstacles and keeps the car in safe zones.
In the tests conducted so far, drivers use remote control to steer a utility vehicle through the obstacle course and monitor through the car’s field view through computers. The system works best for those who trust it. According to Anderson, when the system is installed in regular vehicles, drivers will barely notice it. It will work discreetly in the background and help them safely maneuver the road. This may prove detrimental to drivers – new and experienced, as it diminishes their ability to learn from mistakes and makes them dependent on automated systems. Training humans to become better drivers before working with robotic help is equally important.