Prev: online shopping
Next: change row size in uitable
From: Andrea on 2 Feb 2010 04:35
Hi,

I am creating a controller to mare a rc car follow a path generated by my movement. The problem that I have is that the path generated from the vision system will not be feasible nor will the associated timing law due to the jumps in velocity. Can anyone suggest an appropriate way of filtering this information in real time to provide a smooth trajectory for the vehicle to track.

I am using an extremely accurate external vision system so there is no uncertainty involved.

Many thanks
From: Rune Allnor on 2 Feb 2010 04:40
On 2 Feb, 10:35, "Andrea " <tt...(a)lboro.ac.uk> wrote:
> Hi,
>
> I am creating a controller to mare a rc car follow a path generated by my movement. The problem that I have is that the path generated from the vision system will not be feasible nor will the associated timing law due to the jumps in velocity. Can anyone suggest an appropriate way of filtering this information in real time to provide a smooth trajectory for the vehicle to track.

Right... This is nothing less than the Holy Grail of
contemporary research on unsupervised motion planning.
Don't expect too many plug'n play answers.

> I am using an extremely accurate external vision system so there is no uncertainty involved.

Can't pin-point exactly why, but all of a sudden images of
RMS "Titanic" started flashing through my mind...

Rune
From: Andrea on 2 Feb 2010 05:56


The reason I included the statement that the vision system is external and very accurate is to differentiate my problem from those with significant uncertainty experienced in simultaneous localisation and mapping, normally solved using Kalman filtering. With onboard sensors and uncertainty, the agent may become lost in its own environment whereas with external vision, the agent receives continuous position and velocity in the global axis system.


Andrea
 | 
Pages: 1
Prev: online shopping
Next: change row size in uitable