|
[DPRG] offroad robot exercises
Subject: [DPRG] offroad robot exercises
From: Chris Jang
cjang at ix.netcom.com
Date: Fri Oct 27 21:42:00 CDT 2006
> 1. Drive in a straight line to a distant point and back, with no obstacles.
> 100 feet is a good test. (1000 would be better!)
Hello Everyone,
One thing I would like to contribute is to think of safety. Most amateur
robots are probably small enough and speed limited that they can not do
any damage. But if you have a larger and faster robot, property damage or
collision with people (especially small children) could be bad. I've not
hurt anyone with my robot but am speaking from experience.
jBot is a very mature design. It is essentially safe by design. I think
David once said that it did not make sense to build an autonomous robot
that can move faster than you can follow it. My robot is on the edge as it
is basically the weight and speed of a small child running. And so I've
many more problems operationally.
Here are some of the safety things I consider important for my robot. I
think they do have value for most autonomous machines, though (especially
large and fast moving ones).
1. The most important function is remote kill. This is the only control
that matters. It must be reliable by design. Ideally, it is separate
from the rest of the control system. If a design uses queues or
other state machines in the control system, the kill function must have
priority. Beware of IPC mechanisms and insure that any failure modes
lead to immobility, not speeding out of control (this has happened to
me - it was not exciting but rather horrifying).
2. If remote kill is not available, then the jBot principle of not
outrunning the human chaperone/operator is ok. It's probably better as
by design, there is no need for remote kill. The human operator can
just physically stop the robot manually.
3. Don't be afraid of calling off a test run. If the conditions are wrong
- too many bystanders, pedestrians, vehicles, technical problems, etc
then just scrub. If you are tired and rushed, sick, then don't test.
You won't be thinking clearly and will make mistakes. Don't rush and
push it. That's when bad things are more likely to happen.
I am sorry if this is pedantic.
Chris
More information about the DPRG mailing list
|
