I thought about doing something like that but forming spring steel ...

If you look back in this thread my solution to that was to use push buttons. They hold the bumper out and also provide the on/off switch to signal a hit.

The easiest way to fix a direction (straight or curved) is a tricycle arrangement where the front wheel is both the steering wheel and motorized wheel. It was common in the earlier experiments in making a robot base but I haven't seen it used in the current batch of robots.

https://cyberneticzoo.com/cyberneticanimals/1979-robot-pet-frank-dacosta-american/

Probably if I had a 3d printer I would just use the same bumper solution used by robot vacuum cleaners.

I like the Roomba floor vacuum bumper too.  On the version we have it’s held in place by a leaf spring.  Here’s what I found in Sandin’s book.

I thought about doing something like that but forming spring steel takes a kiln or some way to heat the steel up to 1500 degrees fahrenheit and then you have to shape it into something that fits your bumper design.  I don’t have the tools for that kind of work.

One thought I had was to use a free running swivel wheel with encoders to show if it is rotating and read its orientation to the robot base.

I’ve seen implementations that use a free running wheel to detect whether or not a robot is stuck.

http://schursastrophotography.com/robotics/stasislogic.html  but your idea of using a swivel wheel to indicate pose takes it a step further.

Tom

Looks good.

Probably if I had a 3d printer I would just use the same bumper solution used by robot vacuum cleaners.

One thought I had was to use a free running swivel wheel with encoders to show if it is rotating and read its orientation to the robot base. If it stops rotating then there is contact with an obstacle (even if your motorized wheels are spinning). The angle with the base shows if it is driving straight or at some change in direction.

Your infrared transceivers would have to be integrated into the bumper or have windows in the bumper for them to look through?

     Here is what I came up with using my Prusa MINI+ printer to prototype the layout.  The chassis can hold two N20 type gearmotors, two small ball casters and a QTRX reflectance sensor.

     The bumper base holds one end of each “spring”, three SPDT switches and eventually a voltage divider PCB.  I picked up the tiny SPDT switches at Pololu Robotics in Las Vegas. https://www.pololu.com/product/1528

     Finally the bumper with my improvised “springs”.

     With bumpers that use contact switches I’ve noticed that the lower the switch is mounted on the robot the better is the performance.  I couldn’t  work out how to mount the motors above the bumper mechanism, so I placed the bumper over the motors.  Because of that I’ve played around with enclosing this setup with another platform to help stabilize the floating bumper and it seems to help.

     I always find mechanical bumpers a little fussy.  For example, if the robot runs at a low speed then the bumper needs a low tension in order to close the switches.  Then if the bumper tension is too low it wobbles around and gives false positive readings.  There is also the max operating force required to trip the switch to consider.  This bumper design has the same variables, but using a 3D printer has made it easy to tweak the layout and give it another go.

Tom

Posted by: @zander

@frogandtoad It was different, but competing against IBM was almost impossible. I was only there for a couple years and then moved on. That was early 80's, IBM is a very different company now. I finished up working with Hardware Fault Tolerant computers in the markets segment, stocks etc. Interesting software challenges, everything had to be super fast and converting all kinds of data formats to one modern format was many nights spent poring over assembler output. My system ended up being the fastest but on the day of the big crash 87 nobody could keep up. We did recover first though. One exchange insisted we show transactions in 1/2 of a 64th, NOT 1/128th, but it had to be 0.5 64ths. Now that was weird, and with only a few days lead time.

Hehe, not sure those calculations make a whole lot of sense to me, but I understand what may have been frustrating for you and others!

Cheers

Posted by: @zander

@frogandtoad Just an interesting side note, I left IBM (that's HAL + 1) to work for Hitachi America Limited selling the model 9000 which was the most powerful computer at the time.

Awesome stuff, that would have been a fantastic experience!

I liaised with IBM employees for the last 15 years in the financial industry, troubleshooting networking infrastructure, and overnight scheduling using the Control-M scheduler.

Cheers

Posted by: @thrandell

Posted by: @frogandtoad

Love the vocals on your robot... can you please advise which library you used?

Cheers

Are you planning to make a singing bumper?

 

Tom

Hehe, I actually love singing in my spare time, but nooop... I just like the robot voice you used, as it kind of reminds me of a more modern HAL!

Unless it's a secret, which library did you use? :-)