# Older blog entries for cschur (starting at number 95)

Hi all,

we made great progress this weekend on our Docking Logic program. After determining the range and current requirements for detection with a PNA4602 last week, we next constructed an "Omni Cone" sensor. This is a reflective cone which directs the light from all directions to a single sensor below, and indicates the proximity to the charging base, but not its direction.

It took five pages of hand written mathematics to define how to make such a cone, and how to make one from a flat piece of reflective material starting with a circle with a pie shaped cutout.

Once the cone was made, we place the sensor under it at the calculated distance, and confirmed its optical properties by shining a laser horizontally at the inverted cone, and the beam hit square on to the center of the sensor, from all angles around the cone! Ill of coarse be posting this set of calculations so you can make an omni sensor for your robot as well in the final write up.

Next, we are working with directional sensors, and how to home in with perfect accuracy on the charging beacon.

Write me,

Chris

comets133@yahoo.com

Hi all,

We are proceeding well with the first tests of the huge Docking Logic program. The sensitivity of the PNA4602 is startling to say the least! First we constructed an wave form generator by programming a PIC 12F629 chip to go high 12uS, and low 12us. We used a 20mhz xtal for accuracy. WE got 38.46mhz on the counter. That drives a 2n3904 which drives the IR Led. The range was tested with various in line current limiting resistors.

I first started with a standard 270 ohm, and had a range of over 20 feet to make the PNA device active! So powerful was the beam that the whole room lit up in IR and specular sources were everywhere. Next we put a 50k pot in line, and adjusted to set the distance the device would activate. Much to our suprise, even at 50k, the PNA device would slam on at about a foot. Phenomenal. For our testing, we will now vary the power level to simulate the effect of the beacon and how the robot will react to it at its limits as well as when it is strong.

Happy robots.

PS: did you get your wowee robots from macdonalds yet???

Chris

24 Feb 2007 (updated 24 Feb 2007 at 16:48 UTC) »

Hi all,

First of all, I want to key you in on what I found in Phoenix where I work yesterday - the Mcdonalds Robosapien Adventure meal! If your thing is those crazy humanoid bipeds, the minuature one at Mcdonalds will surely excite you.

Now, onto our project. We got our parts in the mail last night for our PicBot 5 program, Docking Logic. This will be the most extensive, lavishly illustrated project yet in this series, which have been recieved very well so far in the world wide robotics community Im happy to say. HEre is the outline for what we will be researching in "Docking Logic":

1. The Beacon sensor will be the venerable but getting obsolete Panasonic PNA4602, which runs at 38.5 khz. This will not interfere with the IR Prox sensors, which are the IS471 which run at a very different 8 khz. First we will determine the range for various IR LED combinations at the beacon source.

2. Next, we will evaluate using a omni cone reflector on the reciever and determine range and cone configuration.

3. Directional sensor design for homing on the beacon.

4. Robot path tracking from current position to beacon docking. I assume it will be some sort of zig zag S curve.

5. Beacon beam path aquisition. In other words, what to do when we encounter the docking beacon beam.

6. Charging base docking envelope. How accurately can we stop at the beacon in a position ready to charge?

7. STopping at the beacon/charging base in the exact spot without hitting it.

8. Electrical contacts for the determined docking postion envelope.

9. Error correction if docking is unsuccessful. In other words, adding some AI to the docking proceedure so the robot can react more intelligently to mis docking issues.

10. Spin Nav docking. (looking for the beacon with a photocell at the bottom of a tube)

11. Maybe docking with an omni sensor. (the Roomba uses this rather dubious method)

Note that I wont be covering the actual chargin proceedure yet, thats the Picbot 6 project!

Anyway, its going to be the most exciting project yet, and should fill a huge gap in the knowledge base for online tutorials on a subject that is often considered way too difficult to implement in a home robot - self charging.

Write me.

Chris

comets133@yahoo.com

Just got back from a 3week vacation in southern arizona, we were fossil collecting with our friend that flew in from Australia. Anyway, thanks Steve for posting my article.

We are officially kicking off the next project now. The PicBot 5 program, will extensively detail one of the most important aspects of a household robots existance - Docking. This could be for battery charging, reseting wheel encoders to a known reference point, or simpy navigation of a complex area. Needless to say, if your robot cant dock, it cant "feed" and will cease to do its assigned task.

I will be using a new PIC for this project, the 16F876. It has many powerful features, and enable the docking proceedure to be successful.

Later,

Chris

comets133@yahoo.com

30 Jan 2007 (updated 30 Jan 2007 at 03:59 UTC) »

Hi all,

Well, after a few months of working with our stasis sensor bot, weve got what I feel is a pretty decent write up on what is an often overlooked but extremely important home robotics concept. Here is the new article:

http://www.schursastrophotography.com/robotics/stasislogic.h tml

I would be greatly appreciative if one of you could look the article over, and again, if you think the group can benefit from this article, Id like to ask if you could post it in the robots.net main page. Thanks!

Feedback appreciated,

Chris comets133@yahoo.com

Hi All,

An update on the Stasis project. As you know, the stasis detector is the robots last line of defense on keeping from getting stuck in a household enviornment. In its simplest form, it consists of a drag wheel that is pulled around by the robot, that has an optical sensor to determine if the wheel is moving. So your drive wheels may be turning, but if the drag wheel is not, your stuck. The stasis sensor can either be used as a simple sensor input in your main processor, or even better, when used in a priority arbitration architecture or subsumption type architecture, it is the behavior that overlies the bottom behaviors that continue the robot moving toward its goal.

For example, "Random Wander" maybe the lowest level, overlain by "Seek Goal". Both will always keep the robot moving toward its destination, be it the battery charger or to pick up a can or ball. Overlaying these two behaviors with the "Stasis" behavior will be even better. When the robot gets stuck and the bumpers or IR dont see any problem, then after say 3 seconds or so, the stasis behavior kicks in and attempts an escape maeuver to free the robot from say a lamp cord wrapped around the back wheel.

These three behaviors can be overlain by the ones that actually perform a task when the destination is reached, and will of course subsume over the Stasis behavior so that if we are stopped to pick up a ball, the stasis will not kick in and try to escape.

We have experimented with optical stasis sensors too, and it is amusing to put a phototransistor in a short tube and aim it say at a 45 degree angle from straight ahead. What happens is that when the robot is moving, the output of the transistor is a changing brightness, and thus we know we are moving. IF the variations stop, or are very small, we may be stuck. You can also look straight down at the floor and do the same thing, assuming of course that the floor is not featurless!

Such a simple thing - the stasis sensor, but absoulutely essential for the household robot! Dont skimp on this feature, you wont regret it.

Later.

Chris

Hi all,

Weve just started our next project, on robot stasis sensors. As some of you may know, stasis sensors tell the robot when it is stalled, stuck and not moving, or immobile in some other way when its supposed to be moving. A typical sensor might be a wheel that it drags along, and if it stops but the main wheels keep moving, there is a stasis condition.

PicBot IV will explore this concept, and develop a robust stasis sensor system that can be easily incorporated into any home robot. Of course, at the finish of this research project, we will post a very in depth article!

Write me, Id love to hear from you!

Chris comets133@yahoo.com

HI all,

The huge IR Prox Logic article is done and posted. If someone could kindly look it over and if you think its usefull to the robotics community, please post it as an article here, I will be forever in your debt.

http://www.schursastrophotography.com/robotics/irproxlogic.h tml

Chris

Hi All,

We are nearly done with our IR Prox Logic article. Getting a last few graphics done, and It will be ready to post. Some surprises when working in a household enviornment with IR Prox for navigation. We found that tunnel entrapment and small room entrapment to be major issues. With bumpers as you recall, the robots physical diamter is equal to the sensory diameter, and it can usually bounce its way out of a tight space.

Not so with IR prox. The robot is virtually 3x bigger, and can get itself into some fine predicaments in the home. What seems to happen is that a robot will get itself into a small confined area, and then cant get out because the reflectivity in IR on the outgoing trip is higher than the inbound. Thus it cant escape. Long corridor entrapment is also a major issue. A robot can make its way down a narrow hallway then it will at some point start hitting the walls. It can get itself wedged in so that all IR sensors are active at once. That is a difficult one to get out of!

Finally, we did some deep IR photography of a household enviornment at the same wavelength as the robot will see it in IR. The biggest surprise is that wood that is unvarnished is bone white. varnished wood is dark. whew.

Chris

Well, we have finished all the arena testing and evaluation of the PicBot III unit, for exploring the pros and cons of IR Proximity navigation. Now, we are working on the final report, to be shared when it is complete. There were a few surprises with the IR prox sensor array.

First surprise was how much larger the robot becomes when "non contact" IR prox sensors are used. In other words, the virtual size of the robot nearly triples for white walls, and only increases an inch for black obstacles! And to make matters worse, you dont even know what the reflectivity of your obstacle is, so are you a foot, or an inch away? This creates some amusing tunneling entrapment situations. You can get stuck in a confined area that looked dark when you went in, but is white when you turn around to escape - and thus you are quite trapped.

Another surprise was the reflectivity of obstacles in 940nm IR is quite unpredictable. Plants are pure white, so are my black socks! Black poster board is a neutral grey, and black IC foam is totally invisible. To help visualize this, we are adding to the write up an extensive set of images of household objects photographed in deep infrared so you can actually see what the robot will see when it moves about. Pretty cool, ay?

Chris

comets133@yahoo.com

86 older entries...