UPS recently tested a self driving truck. How do self driving trucks work?
In: Engineering
Self driving vehicles are categorized in 5 different modes under the term “autonomous driving”. See Wikipedia article https://en.m.wikipedia.org/wiki/Self-driving_car.
Self driving means normally mode 5, which does not need human interference during the drive. The vehicle doesn’t even need a steering wheel or windows then.
Such vehicles are using lots of technical instruments to be aware of their surroundings like GPS, radar, sonar, cameras etc.
GPS tells the vehicle where it is, while the other technical stuff tells the vehicle what’s going on around it. I.e. other cars on different lanes, traffic signs, pedestrians etc..
All that stuff is processed in the vehicles onboard computer, which decides what the car is actually doing (braking, steering, shifting etc.).
I presume you mean “how do they guide themselves “
Two basic technologies are used: GPS and edge detection ( to put it *very* simply)
GPS uses an accurate clock to measure tiny differences in timing signals from multiple satellites. From this data, using basic trigonometry, you can ascertain your location. ( the more accurate your clock, the smaller the margin of error, and the more accurate your position)
Edge detection. I’m being very broad here. But the vehicle ( autonomous cars do this too) uses various sensors- radar, thermal, optical etc to take a stream of photographs of its surroundings. Using a computer, it checks each photo ( each frame of a video can be treated as a stand alone photo) for certain general shapes, and where they are in relation to the vehicle. There is also a comparison made between the sequential photos to check relative motion of objects- that is, does that object appear to be moving, and in what direction? If it’s getting too close, evasive action is taken.
So, in a nutshell, GPS tells the vehicle where it is, and the sensor system lets it avoid colliding with things.
Combine that with a computer that has a set of street maps, and… there you go.
Simple in concept, a nightmare to actually engineer.
The current wave of self-driving trucks will be limited to highways, interstates and small amounts of in-town driving (Just enough to get them to their dropoff location). AI software and powerful processors have enabled them to become practical and low cost. They use various sensor technologies to make this possible.
**GPS:** This is what determines the vehicles position and when and where they need to turn.
**4G LTE:** Allows the vehicle to keep the map data accurate and to access road condition reports to determine if the current route is passable. For fleet trucks, this will also allow them to stay in contact with HQ. This may also allow multiple trucks from the same fleet to travel in a convoy, though they do not have to use cellular data to do this, they could use different wireless technologies.
**Cameras:** These vehicles will use at least one camera, with one always facing forward. These cameras will look for things like other cars, road lines, traffic cones, traffic signs, and potential obstacles such as road debris and pedestrians. Side cameras are important for detecting nearby cars adjacent to the vehicle. Rear cameras can be used to detect if the vehicle is going to be passed or possibly rear-ended. Front cameras will detect the road lines to help keep the vehicle centered and to navigate turns.
**LIDAR:** This is a special type of camera that can detect the 3D shape of objects. Normal cameras can have difficulty detecting 3D shapes. Usually only one of these sensors is fitted to a self-driving vehicle.
**Ultrasonic:** These types of sensors have been fitted in vehicles for a long time (Look for the tiny circles in the bumpers of cars), most used to detect cars in bling spots and assist in parking. These are useful for detecting the distance between two objects, but it only works in close range. These can assist self driving vehicles by assisting them in parking or checking for nearby vehicles before switching lanes. They can also assist in determining clearance when navigating tight spaces.
**Radar:** Radars have more recently been fitted in many vehicles, they are used to automatically adjust cruise control to the speed of the vehicle being followed, and to detect potential forward collisions. They are used for similar things in self driving vehicles.
Not only UPS, but many companies are relying on LIDAR and GPS. There’s many problems that will arise from using that tech. Once you solve vision (cameras) you will be able to have a full self driving car (level 5 autonomy). Elon Musk explains it really nicely during Autonomy Day for Tesla.
Self-driving trucks follow a pre-determined route (just like the trucks with drivers do) and stop as needed. The idea is that the package-delivery-person can be locating your box(es) while the truck drives to the next stop. This would eliminate the time the UPS truck spends stopped while the driver looks for the box.
Latest Answers