I would be interested in seeing the "tolerance" of the guesses. That is to say, the map on the left showing not only the best guess, but a ghosted image of possible guesses.
The robot is able to navigate with just wheel encoder information and bump contact sensor information. We have developed an algorithm that uses this information to estimate the robot position. This will work with any robot with similar sensors...
What you state is not possible, unless something else is introduced in the environment. Odometry alone plus bumping information is not accurate enough to generate such precise trajectories and localization. Could you please confirm that?
Well, it depends on your definition of 'precise'. The system is precise enough to navigate around a series of rooms - in fact the more complex the room, the better the system works. It is not as good as a laser scanner SLAM approach. The system does require a map of the rooms as a starting point. The algorithm is continually guessing where the robot is based on all the information, usally seems to be accurate to within 30cm.
I would be interested in seeing the "tolerance" of the guesses. That is to say, the map on the left showing not only the best guess, but a ghosted image of possible guesses.
Ultimape 3 weeks ago
really cool, it seems working wery well.
bassanio78 3 years ago
Is the map generated by SLAM? what type of path planning algorithm is it using?
cltw666 3 years ago
is it orienting itself recognizing the objects in the room ? how does it work ???
aneuryzma 4 years ago
The robot is able to navigate with just wheel encoder information and bump contact sensor information. We have developed an algorithm that uses this information to estimate the robot position. This will work with any robot with similar sensors...
merlinlive 4 years ago
What you state is not possible, unless something else is introduced in the environment. Odometry alone plus bumping information is not accurate enough to generate such precise trajectories and localization. Could you please confirm that?
pepevilluela99 3 years ago
Well, it depends on your definition of 'precise'. The system is precise enough to navigate around a series of rooms - in fact the more complex the room, the better the system works. It is not as good as a laser scanner SLAM approach. The system does require a map of the rooms as a starting point. The algorithm is continually guessing where the robot is based on all the information, usally seems to be accurate to within 30cm.
merlinlive 3 years ago
I see. I understand how it works now. Thanks
pepevilluela99 3 years ago