Well it is November again. Which to some of us means a chance of snow. I was sitting around yesterday, looking for something to do. Well I found it. Thinking back a year ago, I remembered, how much my back hurt, from shoveling my walks.
Several years ago I had an idea, What if there was a way to make a machine shovel my walks for me, it could be controlled from my computer, and I watch what was going on from my screen. Well at that time, things like ArduRover simply didn't exist.
Now things do exist that could make this thing a reality. If a CAN (controller area network) system could be interfaced with ArduRover for the control stuff, hydraulics, heavy electric motor controls, and so on, there could be one of these included in everyone's arsenal for countering winter's assault. Well probably not everyone's, but for those fortunate enough to have a six figure income could have one. Well that leaves me out.
This was drawn with SketchUp, animated with SketchyPhysics. I used a video capture program to make the video, so the quality isn't top notch, but good enough for viewers to get the general idea.
Enjoy, hope it gets those gears turning.
Comments
I agree with you 100%. Granted this thing is very impressive, and fast (too fast). Using pneumatic in the winter is a bad idea unless you use nitrogen, or very dry air. Nitrogen is thermally stable. I am sure you have heard the story,
A customer goes to a gas station in December to have his tires filled up. The attendant comes out to assist. the customer says "I don't understand why my tires are low, they were fine in July. The attendant responds, "well you have to fill them up with winter air, the summer air is no good any more".
Anyway back to the topic at hand. In a machine like this, safety is paramount, granted you can't design something so there is no risk at all in operation. Bad things may happen anyway. All a designer can do is his best to make it as safe as possible, for 80% of the time.
I had a feeling some of these existed, at least in the form of remote control plows. And that is an essential stepping stone to the more interesting and challenging problems of adding some autonomy. Further, there is, without a doubt, a lot of great code and experience that might come out of something like this which would benefit all of the ArduPilot projects, not just ArduRover. Rock on. I like the hazard lights, and worry that the operator in the video is not taking enough safety precautions by rolling something like that in those locations via video alone. Too much opportunity for a car or truck to hit it, or a kid to bike or fall in front or behind it... that much power, it could be fatal for a kid, or pet, and very costly for a driver who is backing out of their drive way at the same time... (since it is low enough, you could easily miss it as a distracted backing driver.)
By the way gents, my idea already exists, in a different form. here is a vid of a unit doing it's job.
Mike, the kill switches (4) would be on the unit, one on each corner, already thought of that. At first the safety operator would be behind the unit carrying a switch attached to a tether, attached to the machine, not unlike the dead-mans switch, in a train locomotive. In the case of the train locomotive, one of the train crew has to push a switch every two minutes, while the train is moving or the locomotive will start a shut down sequence.
In a commercial application of my machine, there would less likely be an issue. In a residential application the scenario changes.
It would be easy enough to set up proximity switches of some type, for activation of the kill system, one in front, one on the rear, the appropriate sensor activates, dependent on the direction of travel, the front sensor when travelling forward, and the rear sensor, when travelling in reverse. A passive infrared sensor presence sensor would not work in an outdoor environment, in the winter, a person, properly attired would "look" the same as the environment, to the sensor. However an ultrasonic, or microwave device wouldn't care. I am thinking microwave might be the ticket.
Funny you should be thinking about this. We've been thinking about ArduRover a lot, too. See my post here.
sorry.... just idle thinking, I don't mean to sound like I'm stating any type of requirements, just a mental exercise...
I think you want on-vehicle safety kill switches, accessible from any side, and it seems to me a good practice to use a deadman's switch on the TX that you use for the heartbeat. This would give the same sort of safety feature used for most push devices like a self-propelled lawn mower, right?
If those things were done (so that anyone could run up to the device and stop it, and so that if the operator passes out or is run over) then the main difference in safety between direct operation and computer controlled operation would be that the operator 1) might be less attentive and 2) is generally going to have a different/more obstructed perspective, since he is not "behind" the device, but rather standing at a static point some distance away, with a blind spot on the opposite side of the device.
Hi Mike, I backed off my original concept, a robotic snow blower I decided it would be a huge liability, wouldn't take much to chew up and spit out the neighbor's Chihuahua, or a curious child. A blade is a more passive choice, but still has the potential for causing harm.
I didn't put the E-stops on this model, but I thought about them, one on each corner, when activated, the main power would be stopped, either batteries or an ICE.
Now comes the issue of a remote E-stop, not an easy task. There are some commercial radio control set-ups that have this feature and they are OSHA tested and approved. The only way to do this remotely is with a separate radio rig, that would not be prone to interference, and a heartbeat check, that is, a signal sent from the transmitter every 100 milliseconds or so. If the receiver doesn't get that signal, at the specified time interval, it shuts the machine down NOW! Still there remains a large liability.
So in a prototype stage a human safety operator would always have to be in close proximity, just in case, but even that is better than a backache!
Thanks for you comment!
In fact, the kind of work done to support a project like that would be of value to all ArduRovers, I think. Detecting wheel slip, managing/detecting road grade, an uneven surface, more complex pathfinding required for ground navigation (think zamboni, bulldozer, lawnmower, or snow clearing... not that I'd want to do either of the last two, for safety reasons... speaking of which, need reliable kill switches) many specialized conditions that deserve special handling.
There are quite a few robot bulldozers teams used out there to build lakes (with lasers!) and other landscaping projects. I could certainly see an ArduPilotMega used as a controller for something like that.