[chbot] Pushing a trolley with a robot

Charles Manning cdhmanning at gmail.com
Wed Aug 20 20:44:57 BST 2014


I spent a lot of time at Trimble figuring out steering algorithms for
tractors etc. http://www.trimble.com/agriculture/ez-steer.aspx?dtID=overview

One of the hardest problems I solved was that of steering an implement
on a movable coupling. This involved steering both the coupling and
the tractor to keep both in sync. It is, in essence, the same problem
as you are trying to solve here (just with the extra wrinkle of
avoiding jackknifing).


There are really two parts to this:

The coupling/steering mechanism mechanicals + feedback sensors.

The control software. Steering control is hard enough to just keep
things on line and stable when you're driving forward, when you also
have to "pre-steer" to push something along, it gets even harder.

If you are an inexperience and unskilled  trailer-reverser such as
myself, you will know how easy it is to jackknife etc.

The only way I can think of making this task mush easier is to make
the hitch moveable like your lead screw idea. That is used in movable
hitches on tractors to keep the implement in line when the tractor is
offline.

The big problem you are trying to avoid is keeping the lead screw
thing from running out of room and then being unable to avoid a
jackknife.

The way to do this is to use the lead screw to actually control the
steering of the whole machine. The trailing part is then steered to
keep the position of the lead screw near centre.

ie. you have two PID loops:

The steering PID loop (fast) that uses the offline of the front
trolley as input and controls the lead screw. It controls to minimise
offline. (Approximately, there are better ways of doing this to
prevent oscillation etc)

The correction PID loop (slow) that takes the position of the lead
screw as input. It then steers the truck to recentre the screw.

The reason the correction PID must be slower is that we don't want its
changes to propagate through to the steering. The screw PID must be
able to wind fast enough to swallow up those changes.

Hope that helps.




On Wed, Aug 20, 2014 at 5:56 PM, Andy Gardner <ceo at andygardner.com> wrote:
>
> How about having the trailer coupling mounted forward of the trailer wheels (above the scoop system), then a bar coming out the front of the truck, up higher than the height of the trailer going over the top of the trailer, then attaching to the coupling.
>
> If you get what I mean. It might look a little silly, but it would work and would be a lot easier, software-wise.
>
>
> On 20/08/2014, at 4:40 PM, Robin Gilks wrote:
>
>> Greetings all
>>
>> I've hit an interesting problem with my walnut harvester.
>>
>> The actual nut picker-upper (the trolley in the diagram) is a push device,
>> obviously you can't pull it along as the wheels of the drive unit will
>> either crush the nuts to be harvested or push them into the grass out of
>> reach.
>
>
> _______________________________________________
> Chchrobotics mailing list Chchrobotics at lists.linuxnut.co.nz
> http://lists.ourshack.com/mailman/listinfo/chchrobotics
> Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
> Web site: http://kiwibots.org
> Meetings 3rd Monday each month.
>
> When replying, please edit your Subject line to reflect new content.



More information about the Chchrobotics mailing list