February 28, 2014
Starting this blog as a place for me to throw down my thoughts about options for my Service Droid, VADER:
At times where I'm anxious to work on Vader but can't, atleast I can blog about it! Right now is a good example: my 2 month old daughter is asleep in my arms. I may be having to type with one hand but the progress feels good!
First off, I'm considering a tweak to the acronym."Very Advanced" is a little over the top. Maybe "Virtually Advanced" or "Very Amateur".
Second, I keep asking myself "What is Vader's mission"?
I have a theme, but not certain of the details. In general I'd like him to identifiy a target given certain characteristics, retreive the target and return it to the starting point. For example, (Scenario 1) if I set a coke can in a square room he would have to scan the room for something approx 3" wide and 6" tall. After identifying the target and it's location he would navigate to to the can, align an arm to the can and pick it up. After picking up the can he would then return the starting location.
Another variation on this theme (Scenario 2) is that a remote user (me) uses a wireless camera and a range sensor in Vader's head to identify the target. Then using the distance and the bearing derived from the pan/tilt angles from the head, he would navigate to the location of the target. He completes the task again by picking up the can and returning to the starting point.
The primary difference between the two scenarios is the first requires additional scanning capability to find the object.
Now, can I be so bold too as to assume that there could be objects between Vader and the target? Good question. From the vantage point of Vader's head I could locate the target but do it while looking over smaller objects. Now when he attempts to go from point A to B he will have to detect and navigate around obstacles and regain the correct heading to the target. If he is in charge of finding the target himself he could simply wander around while avoiding hitting stuff and scan each obstacle to see if it approximates the target dimensions. If it doesn't then he simple turns past it and keeps wandering.
I've come to the conclusion that I'll likely have sensors on the head as well as on the front end of the Rover 5 Chassis. If for any reason this config will be useful for object avoidance since sensors on the head can overlook them.
So now I get to the meat of the questions I currently am chewing on.
1) Which sensors would be best suited for Vader's mission?
2) How and Where should they be used?
I could work to Scenario 2 and evolve to Scenario 1 but only if the answers to these two questions are not mutualy exclusive in both scenarios.
So far I have done some level of research on:
1) SR04 Ultrasonic scensors -
The good ol' cheap standby. My baseline design uses two of them in Vader's head. I have thought of adding a third at the front end of the Rovoer 5 base. The concerns I have with relying on these sensors are the slow refresh rate and the large 'spot' size is sees. Trying to distinquish a 3" wide object or smaller at some feet away may be difficult with these sensors
2) Sharp IR Sensors
I'm really tempted by this new device. It won't be available until June, but it has the advantage of being a 5V device capable of measuring up to 40m (yes, meters). It has an update frequency of up to 100hz with a 1.5 Degree beam width. I would say this is my leading interest right now. The crowdsourcing event has about 3 weeks left as of 2/28/14.
4) Opensource LIDAR
Really cool device but it requres 12V and has a max update rate of 50hz. Also uses a real laser which comes with safety concerns. I have two dogs and 4 young kids in my house....
5) Neato Scanning LIDAR
I love the results this thing can provide and I'll probably get one someday. I don't think however it will work for Vader is rather large- about 6"x5", it draws 300 milliamps and uses a real laser as well.
6) Maxbotics Ultrasonics
I need to read more about these to be honest. Maxbotics has some real improvements in these ultrasonic devices over the SR04.
I need to read more about the results people have gotten from this device. Chris the Carpenter sells these at Rocket Brand Studios.
Now, back to the Scenarios.
OddBot has been working on a concept of using infrared LEDs to enable one of the SDs arms to scan an object and align with it. Vader will need this capability too. Another idea I was considering is using row of IR LEDs and photo transistors along the front of the Rover 5 that would flash at a potential target. Using the returned voltages from the phototransistors I might be able to tell how wide an object is, or attleast the maximum width of the object. For example if I have 10 IR sensors spaced on 0.5" centers and sensor 1 and 10 return much less voltage than 2-8 then I can say the object is likely less than 5" wide. If sensors 1,2, 9 and 10 return much less than the rest then teh object is likely less than 4" wide and so on. Kind of a crude method but it's in my toolbox....
More to come....