Give a robot the ability to pick up behavioral cues -- looks like the guy ahead of me is about to turn left -- and it'll more accurately follow you around.
Robot controller software that adds behavioral cues to the follower robot's estimation of where the leader is going helps the robot follow it. Behavioral cues could include the direction a robot leader is looking.
The software could make robots work more effectively and safely with humans.
Research paper:
Following Controller for Autonomous Mobile Robots Using Behavioral Cues
IEEE Transactions on Industrial Electronics, August 2008
Researchers' homepage:
Robotics, Autonomous Systems, and Controls Laboratory, University of California, Davis
Navigation is nothing more than plotting an efficient route from point A to point B. Fundamentally, robot navigation includes just two things: the ability to move and a means to determine whether or not the goal has been reached. The trick is finding the most efficient way to reach a destination. There are several aspects to this seemingly simple problem and several ways to solve it.
In the age of sailing, navigating meant finding the ship’s position using the stars, charting the position on a map, drawing a line from present position to destination, and deriving the compass heading for the ship to follow. Today’s ship navigation uses Global Positioning System readings rather than the stars and electronic maps rather than paper ones, but the principle is the same.
This doesn’t work in all situations, however. An autonomous mobile robot might not have a map of its environment handy or be able to determine its position, but it still requires a means to get where it’s going. In instances like these robots need to be able to navigate more like people, who can read maps (“you are here”), follow a heading (“go four blocks north”), recognize landmarks (“turn left at the store with the big purple sign”) and use waypoints (“the meeting is at 125 Main St., 4th floor, room 10”).
There are three major types of robot navigation.
Big picture
A robot that uses map navigation must have a global representation of its environment. The robot makes some kind of measurement to find its position, and plots a course to its destination. The robot has knowledge of all the locations in the environment and how they are related to each other, and knowledge of its own relationship to the locations. If the robot is initially given its position on the map, it doesn’t need any information about its surroundings to reach a destination.
Bread crumbs
A robot that uses waypoint navigation follows a sequence of recognizable landmarks to reach a destination. The robot is aware of locations beyond its sensor range, but does not know the relationships among the locations. It finds its way from one landmark to the next using local navigation techniques. Robots can also use waypoint navigation to build maps for subsequent map navigation. When multiple sets of waypoints can be used, the robot must be able to plan a route.
How it looks from here
A robot that uses local navigation taps sensor data to determine its position relative to observable landmarks and compares this to the destination’s position relative to the same landmarks. The robot changes its position until it matches the destination. Local navigation requires robots to be able to recognize destinations, aim for them, and hold a course.
Robot controller software that adds behavioral cues to the follower robot's estimation of where the leader is going helps the robot follow it. Behavioral cues could include the direction a robot leader is looking.
The software could make robots work more effectively and safely with humans.
Research paper:
Following Controller for Autonomous Mobile Robots Using Behavioral Cues
IEEE Transactions on Industrial Electronics, August 2008
Researchers' homepage:
Robotics, Autonomous Systems, and Controls Laboratory, University of California, Davis
Navigation is nothing more than plotting an efficient route from point A to point B. Fundamentally, robot navigation includes just two things: the ability to move and a means to determine whether or not the goal has been reached. The trick is finding the most efficient way to reach a destination. There are several aspects to this seemingly simple problem and several ways to solve it.
In the age of sailing, navigating meant finding the ship’s position using the stars, charting the position on a map, drawing a line from present position to destination, and deriving the compass heading for the ship to follow. Today’s ship navigation uses Global Positioning System readings rather than the stars and electronic maps rather than paper ones, but the principle is the same.
This doesn’t work in all situations, however. An autonomous mobile robot might not have a map of its environment handy or be able to determine its position, but it still requires a means to get where it’s going. In instances like these robots need to be able to navigate more like people, who can read maps (“you are here”), follow a heading (“go four blocks north”), recognize landmarks (“turn left at the store with the big purple sign”) and use waypoints (“the meeting is at 125 Main St., 4th floor, room 10”).
There are three major types of robot navigation.
Big picture
A robot that uses map navigation must have a global representation of its environment. The robot makes some kind of measurement to find its position, and plots a course to its destination. The robot has knowledge of all the locations in the environment and how they are related to each other, and knowledge of its own relationship to the locations. If the robot is initially given its position on the map, it doesn’t need any information about its surroundings to reach a destination.
Bread crumbs
A robot that uses waypoint navigation follows a sequence of recognizable landmarks to reach a destination. The robot is aware of locations beyond its sensor range, but does not know the relationships among the locations. It finds its way from one landmark to the next using local navigation techniques. Robots can also use waypoint navigation to build maps for subsequent map navigation. When multiple sets of waypoints can be used, the robot must be able to plan a route.
How it looks from here
A robot that uses local navigation taps sensor data to determine its position relative to observable landmarks and compares this to the destination’s position relative to the same landmarks. The robot changes its position until it matches the destination. Local navigation requires robots to be able to recognize destinations, aim for them, and hold a course.
0 comments:
Post a Comment