4-Dimensional Location Mapping -- Cloudlet Computing & Communications

How do we create and maintain a local Map? Let's start with 3-Dimensions and then add time.

Assume each Autonomous Vehicle has its own sensors to help determine where it is and what's around it. Sensors can include optical and other devices designed to identify and locate nearby objects. (I’ll talk more about various types of sensors in future posts, including some ideas for a 360° Adjustable Resolution, Stereo-Optics System.)

Thus each Autonomous Vehicle collects its own information about the surroundings with its sensors. This provides a line-of-sight point-of-view, and is limited by the ability to resolve and identify objects.

Assume each Autonomous Vehicle also has a communication system to gather and share information.

I propose that each Autonomous Vehicle share its information with nearby Autonomous Vehicles. Then each vehicle uses the collective information to calculate a local map. The vehicles then share the resulting map. Each vehicle then verifies that the collective map is consistent with its own information, and if there is a problem, it shares the details of the problem, and the group attempts to resolve the inconsistencies.

Once this process is complete, each Autonomous Vehicle now has a map of the local environment more extensive than its own point-of-view. 

A vehicle entering the local area can immediately gain access to the map and compare with its own data.

Cloudlet Computing & Communications is my name for this system. Because the information is available to everyone, and processing is distributed, it’s like “cloud computing”, but because this is only local information, I propose the diminutive term “Cloudlet Computing & Communications”.

 (The details of the protocols and procedures I have developed for forming, entering, and leaving a Cloudlet, and sharing information and processing, are beyond the scope of this blog.)

Rather than just having your own line-of-sight perspective, your vehicle now has a comprehensive local map of vehicles and other objects in the vicinity. The data are based on multiple observations, which improves accuracy. And the data have been verified from the different observations so that errors due faulty equipment are immediately identified and corrected.

While this process can identify objects including walls, doorways, rooms, furniture, stairways, and even electrical outlets, additional information is needed, for example, this is the Candlelight Dining Room at Charlestown or this is Main Street. Thus assume that this information is either loaded as conventional maps and floor plans, or people enter the information as they use the Autonomous Vehicles.

By tracking object locations as they change over time, each vehicle could predict where each object will be in the future. As I discussed in the previous Post, this approach has a number of shortcomings.

Cloudlet Computing & Communications allows each Autonomous Vehicle to share not only its location, but its current velocity and acceleration, plus it’s Planned Path. Thus the concept of a simple map expands to our 4-Dimensional Map, which is not simply predictive but includes the Planned Paths of all the vehicles.

A simple example of the value of this shared Planned Path information is preventing rear-end collisions and chain-reaction collisions, some of the most dangerous types of collisions. When a car slows down or stops on a highway, the driver behind has to notice and then hit the brakes, a process that can take hundreds of milliseconds, even assuming the driver is paying attention. Each successive driver has to see the car in front slow, decide to slow down, and start slowing before the driver behind can start their own decision process, and this is exacerbated by large vehicles which block the view of vehicles farther up front. When cars are following closely the cumulative delay means that even if the first few cars stop in time, later cars will collide.

When I was 10 years old my father stopped on a highway, 3 cars back from the original stopped car. The car behind us stopped, but the next car back rammed the one behind us, which then rammed into us. It took over an hour for the police to clear the mess. Fortunately no one was injured, but it took us a couple hours and many skinned knuckles to finally get our luggage out of the truck.

With Cloudlet Computing & Communications, the first vehicle to slow or stop would notify all vehicles in the Cloudlet in milliseconds, rather than the seconds of cumulative delay without the communication.

Cloudlet Computing & Communications will play several roles in our Autonomous Transportation system beyond forming the 4-Dimensional Maps. Next I’ll talk about scheduling and planning in our 4-Dimensional Map.