Navigation and 4-Dimension Global Maps, Part 1

To navigate, either in a vehicle or on your own, you need to know both where you are going and what is around you. We use maps to help us decide where we are going and what we are likely to encounter along the way. We use our senses to determine our surroundings, often augmented by various technologies for things beyond our senses: more distant, more accurate, and more detailed.

Unfortunately, it is challenging to convey the information from our senses to someone else.

Maps are efficient and effective methods to represent information about what is around us. Paper maps have been used for centuries, but more recently electronic maps are replacing paper; for example, we use maps in GPS for navigation, on roads, trails, water, and even in the air.

I’ve been led astray several times by incorrect GPS maps, so we need to do better. Also these maps are only an approximation of what is around us and what we will encounter. Even aside from errors, maps don’t contain information about important details like pot holes and other obstructions like parked cars, and certainly not moving objects like people, cars, deer, etc. Also typically maps aren’t available for the insides of buildings, and even detailed floor plans don’t show furniture, wet floors, and other information essential for safe and successful navigation.

Previously, I’ve described why your sensors aren’t sufficient to guide you 4-DimensionalGlobal Map and challenges of some current approaches to navigation Challenges of Car-Based, Situation-Oriented Approach.

I’ve already talked about the powerful mapping systems already available in MMOGs  4-Dimensional Global Maps –Economic & Technical Feasibility, so we know shared maps based on many different peoples’/systems’ are feasible and widely used.

I’m proposing expanding to a new level of mapping, which I call 4-Dimension Global Maps, to meet the navigation needs of the 21^st century, both for Autonomous Vehicles and for people.

Purposes of the 4-DGM:

• Navigation
• Historical data on past activities, both aggregate (e.g., traffic volumes) and possibly individual (what road was I on last Saturday night at 9 pm)
• Future information, which may be projections based on historical data (e.g., usually about 12 cars/min pass this intersection between 8 and  9 am on a weekday and the parking spaces in this block are 90% full, or I usually go to the Starbucks on Main St. at 8:42 each weekday morning J), but also can include, reservations (9 pm at the restaurant for 4 people), general plans (e.g., the traffic light at Main and 1^st Streets runs 90 seconds on and 90 seconds off), and specific plans (e.g., there is a parade scheduled on Main St. Monday morning 9-11 am; and I’m taking my daughter fishing Monday morning, so I will be at Starbucks around 6:15 am). 

How do we form the 4-Dimension Global Maps?

We group information contained in maps by the size of the features and the rate of change. Thus buildings, roads, sidewalks, walls, trees, and large bodies of water (except perhaps for their precise boundaries), and other relatively unchanging items are static elements. At the other extreme we have ephemeral elements, such as the location of specific people, animals, and moving vehicles, as the most dynamic. In between these two extremes we have different rates of change.

A key feature of electronic maps is the ability to zoom in and out, covering a wide area with reduced detail or a smaller area with greater detail. The information in the less detailed maps changes relatively infrequently because cities and highways don’t move often J. Thus we will call these “Static Maps” – even though they may change occasionally. These Static Maps are the framework for the more detailed, more dynamic map information. The Static Maps provide the basic information that a local map can incrementally add detail to. They also provide continuity across boundaries of locally perceived areas. Note Static Map features can change, and we need a process for passing the change information from sensors and other sources, verifying the changes, and distributing the changes to the many copies of Static Maps.

We form local maps from a combination of “static maps” and “dynamic local information”.

As we zoom in from the Static Map of the largest area of interest to an Autonomous Object, e.g., US and Canada, the amount of information increases, along with the likelihood that the information will change in time. As with current GPS maps, an Autonomous Object typically already contains the most basic Static Maps for its area of interest, e.g., to include the area most frequently traveled around the base of the Autonomous Object. It will request more detailed Static Map information in preparation for a specific trip.

Note it is counterproductive to request dynamic information until it is immediately needed: because of the high rate of change, it would be obsolete before you need it and you would just need to request it again.

As we zoom in, items may by more or less static, for example, a massive sofa will be quite static, while the chairs around a dining table move many times a day. A parked car may only stay in place for a few minutes, or it could be in the same place for days. The boundaries of tidal water change continuously with a typical frequency of twice a day.

To accommodate this variability we use the technique of a baseline and incremental updates. The static information, at a given level of detail, will be included in the baseline, and changing data will only need to be sent when a change occurs. This technique is used when landing at an airport, the pilot gives the code for the announced local information, including weather, flight conditions, and special situations, so the tower knows if the pilot needs any additional information.

As you prepare to enter a local area, you already have a baseline of some level of detail of static data. You can state the characteristics of your baseline, and the level of detail you need, and then request an update of additional detail for static information plus the dynamic information at the requested level of detail.

For example, you may already know to follow Main St., but you don’t know how wide the street is (more detailed static information), that there is a 3’ diameter pothole 2’ from the right curb (semi-dynamic information), and the positions and dynamics of all the moving vehicles near your path (dynamic information).

Here are some examples of levels of detail:

• If you are going to go through a door to enter a room, you don’t need to know what’s beyond where you will turn off, unless it might move into your path
• If you are trying to avoid an object, e.g., a table or a pothole, all you need is its outline and position, and the degree of accuracy of the measurements depends on how close you need to come; but if you are likely to hit the pothole you might need to know its depth profile to know where to go through it
• If you are trying to maneuver to sit at a table you need to know down to an inch or so the placement of the legs and their dimensions, the height of the clearance under the table, etc.
• A “car” wouldn’t need the detail of the inside of the room, or the configuration of the heliport on the roof J

•