• Tidak ada hasil yang ditemukan

A practical starting point and “a priori” knowledge

Dalam dokumen Designing Autonomous Mobile Robots (Halaman 132-136)

Disadvantages of goal seeking

Goal seeking is elegant and interesting, but it assumes that the only reason for choosing a route is to avoid obstacles. In most applications, there are other reasons for choos- ing a particular route as well. For example, a security robot’s route is planned to assure that the robot performs surveillance on all important areas. In other cases, there may be traffic in the area that requires rules of the road to be followed.

As mentioned earlier, the major disadvantage of goal seeking is that it takes routing decisions out of the hands of the system programmer. Therefore, if a robot is to patrol a building in a specific manner using goal seeking, then it must be given a sequence of goals that force it to cover the areas desired.

We will then begin by telling our robot where we would like it to go, and what we would like it to do. This is no different than what we might do for a new human worker, except that our means of communicating this information will be a bit different.

We will also help the robot with the selection of features it should use to navigate. We know which objects in its environment are permanent and which are not, so a little help from us can greatly improve the robot’s performance. If we find a method that allows the robot to use its sensors to add navigation features of its own, our architec- ture should make it possible to add this capability at a later time.

Lateral Uncertainty

Obstacle

Obstacle Robot path

Danger levels

Figure 7.6. Topological map of force fields between obstacles

Live Reckoning

8

C H A P T E R

It is ironic that the most central precept of autonomous navigation is often viewed disparagingly, especially by laypersons. Perhaps this is because the term dead reckon- ing was adopted to describe this precept, and most people have a preformed view of what this entails. The very name dead reckoning seems to imply doomed guesswork.

At best, the term conjures up images of Charles Lindbergh flying across the Atlantic Ocean with little more than a compass heading for navigation.

If we announce that our robot has excellent dead reckoning, many in our audience will think this means that our robot has no true navigation at all! At the very least, they will tend to think that our robot, like Lindbergh, travels long distances between updates of its position. For the purposes of this discussion, we will surrender to this view, and use the new term live reckoning to describe what we want to accomplish. To under-stand live reckoning, we should first discuss dead reckoning a bit.

The truth is that dead reckoning was invented by nature long before man coined the term, and this is just the first of many of nature’s techniques that will be useful to us.

One of the Zen aspects of mobile robot design is the fact that we will continually come to these same concepts whether we attempt to copy nature or not!

In mammals, dead reckoning is accomplished primarily by signals from the inner ear.

We may think we find our way across a room by visual navigation, but the fact is that without our inner ear we couldn’t even stand up and walk. Other senses, such as the feeling of pressure on the soles of our feet, play into our navigation next, and finally our vision is used to correct these estimates.

Inertial navigation uses accelerometers and gyroscopes as inputs to its calculations, but is in the end a form of dead reckoning. On the other hand, if the position and

orientation estimate of an inertial navigation system is being regularly updated with beacon or GPS position fixes, then it becomes live reckoning. The distinction be- tween the two is therefore largely a matter of frequency, but it is an important distinction.

To avoid the confusion we will avoid the term dead reckoning. The process of regis- tering relative motion on a continual basis will be called odometry instead.

Relative vs. absolute

At this point, we must differentiate between relative navigation information and absolute information. Odometry is relative information that tells us how much we have moved, but not where we are. If we know where we started, then we can add the relative movement to the starting position and obtain an estimate of our current position.

Odometry gives us very fast feedback, but because of accumulated errors, it is subject to drift. Other navigation techniques will be needed to correct the odometry estimate periodically. Some of these may correct all axes, and some may correct one or two.

These principles have been used for much of recorded history. Late in the Battle of Britain in World War II, the Luftwaffe decided to switch from day bombing to night bombing to reduce their rather severe losses from Royal Air Force fighters. Since England was under a strict blackout, there were no natural landmarks from which German pilots could correct their dead reckoning navigation, especially on cloudy or moonless nights.

Thus, the Luftwaffe developed a ground-based beam system code named “Crooked Leg.” The bombers “rode” a primary pair of beams that provided an indication showing whether they were straying to the left or right of their intended path to the target.

On the left the pilot would hear Morse code dots, and on the right dashes. In the middle, these signals would combine to provide a continuous tone. A second beam pair intersected the first just before the target at the IP (initial point), telling the crew to prepare to drop their bombs1.

1 The British quickly detected the beam system and added a signal of their own to cause the indicated path to be deflected off of the target. The history of WWII is rife with fascinating examples of technologies and counter measures being developed in rapid succession.

In navigation terms, the first beams gave the pilot absolute information about the air- craft’s lateral position with respect to its intended path, but told nothing about the longitudinal position along the path. The longitudinal position was estimated en route from air speed and wind reports. The second beams provided the longitudinal correc- tion or longitudinal fix. These basic concepts are the same as those of mobile robot navigation, which in turn are the same as those used by animals for navigation.

One of the very few absolute navigational references available is the GPS (Global Positioning System). This system provides very accurate absolute position informa- tion by triangulating satellite signals. Best of all, it requires only an inexpensive GPS receiver.

Left

Right Left

Right

IP

Dalam dokumen Designing Autonomous Mobile Robots (Halaman 132-136)