Direction location - blind kayaker tactile guidance

JRiley

Joined Mar 21, 2016
5
Hi,
Looking for direction location ideas.

Our national organization, Team River Runner teaches veterans with disabilities how to kayak. This project is related to working with blind/visually impaired. Traditionally, we've used an audible "On Me" guidance chant that allows them to home-in on our voice. However, some of our participants are also deaf and the audible method does not work - or work well with distance or in wind (not to mention wearing out the guide's voice). In a former life, I was a 2nd class FCC RF technician and am now a software engineer. Recently, I developed a tactile solution using radio guidance with XBee transceivers. 4-way directional commands (forward, back, left, right) are transmitted from a guide radio mounted on a kayak paddle using a small 4-button thumb control. RF direction packets are received by an XBee radio worn on the blind kayaker's life vest and commands are translated to four tactile vibration motors embedded within the vest. The system works well to 100 yards, is lightweight, waterproof and durable.

I'd like to kick it up a notch and use RF direction finding to allow a blind kayaker to home-in on a guide traveling ahead of him. I envision the guide boat carrying either an RF beacon or an RFID tag. For the blind kayaker, a direction-location receiver inside of a waterproof otter box fixed on boat deck or fixed inside the kayak. Using something like antenna switching with yagi's (or other method) the circuit would determine the strongest direction of the guide signal and use the current life vest radio interface to pulse appropriate vibration motors in the vest. The box would be fixed and oriented to the bow of the boat.The functional range should be at least 30 feet. Direction location is the primary goal but if I can also use time-of-flight or RSSI strength for relative distance to the guide and regulate the vibrator strength, all the better.

Concept document is attached. In this document BEACON MODE is what I am researching. DIRECT MODE is functional.

Can you recommend a combination of circuits/products that might work for this? I toyed with using ultrasonic but it seems range is very limited for inexpensive units. I also looked into infrared (IR) sensors but understand they do not work well in sunlight. Environment is outdoors and over fresh or salt water. Would like to avoid messing with processor or mobile phone/GPS. The simpler the circuit and setup, the better.

Much appreciated.
Jim

Attachments

• 240.1 KB Views: 14

Bernard

Joined Aug 7, 2008
5,351
This is a well written worthwhile project. My limited experience with RF direction finding dates to WW2 & required rotating or oscillating antenna. Might give modulated IR another look.

AnalogKid

Joined Aug 1, 2013
8,251

Receiver: For coarse indications, 5 photodiodes, or even simple IR remote control parts (TSOPxxxx) in a semicircle with funnel shaped blinders that partition the forward 180 degrees into 5 directions: hard left, shallow left, straight ahead, shallow right, hard right. 2 rear facing receivers let the boater know he's pointed backwards and which way is the shorter arc back to forwards. Simple logic to convert the 5 forward signals to the four transducers. For example, hard left versus shallow left could be pulsing versus continuous left motor. No microprocessors (sorry, PICboys), no firmware, no RF. A circular, waterproof, IR transparent enclosure is someone else's problem. Somewhere around a small Frisbee in diameter, but the larger the diameter, the less overlap among the direction signals.

Transmitter: Modulation makes it sun-proof, 30 feet is nothing some D cells or a small SLA battery can't handle, the emitter can be a semi-circle string of LEDs in series for even 180 degree radiation. AND, as a Springtime present to the crazies on AAC, the transmitter can be clocked by ... wait for it ... a 555 ! ! ! ! ! ! !

ak

Last edited:

JRiley

Joined Mar 21, 2016
5
Thanks!! Will explore modulated IR more and start breadboarding. Found a couple of example circuits indicated greater range and the modulation reduces sunlight problems.

Bernard

Joined Aug 7, 2008
5,351
40 m range listed with VISHAY TSOP6238TR, IR receiver module ( Electronic Goldmine, A 20419, $.95 ), using IR LED TSAL6200 @ 250 mA, 38 kHz. Looks like there are possibilities. AnalogKid Joined Aug 1, 2013 8,251 The more I think about the light paths, the more I think that the receiver discrimination logic could be better with a uC. For example, if the boat is pointed straight at the transmitter, then the center sensor is illuminated directly by the transmitter. But even though the left and right sensors can't see the transmitter directly, IR energy can hit the outside wall of each of the side tunnels and bounce down to those receivers. So you if you call the radiation at the center sensor 100%, the two side sensors might see 20%. This is not a hard problem to solve in all discrete circuitry, but it might be more simple to sample all 5 sensors and sort it out in firmware - IF you already are a PIC or Arduino wiz. To do this without a uC, there are two approaches. With a fully integrated IR receiver, you get only a go/no-go signal. With 5 sensors there are 32 possible combinations of receiver signals. These can be decoded with the equivalent of a lookup table - two 4line-to-16line decoders and the outputs gated together to drive the vibrator motors. If you do a discrete receiver you get a signal strength analog output. These can peak-sum into a self-adapting threshold detector that always picks the strongest signal, which drives the motors directly. A uC can implement the logic for either approach. ak ErnieM Joined Apr 24, 2011 8,011 I thought RF DF cannot resolve an absolute direction. Would not an RF solution leave open the problem of knowing what you are homing in on is upstream or downstream? ErnieM Joined Apr 24, 2011 8,011 An IR receiver could be no more complicated that a little hand held unit you wave around till it sees the boat to home in on, then it either lights beeps or shakes. The receiver would not need anything more complicated than a TV IR receiver looking for a 40KHz signal, then a transistor or two to drive the led or buzzer. Thread Starter JRiley Joined Mar 21, 2016 5 I see your point using a uC to poll all 5 IR sensors and select the one with the highest received radiation to determine the primary vibrator to activate. I could then decide if a secondary vibrator should activate using some threshold % of the primary level. This experimenting could easily be done programmatically with an arduino. I could also experiment with regulating the poll rate of all 5 sensors as the fewer pulses to the vest vibrators through the xBees, the better conservation of battery power in the vest. For concentrating the IR light, I’m wondering if placing each of the LED’s inside of a flashlight reflector cone (eBay /$1 ea) would help. For that matter, perhaps also the IR receiver modules? For the physical layout and enclosure I'll have a better idea after breadboarding and experimenting.

You've all given me some great ideas to get this rolling. Thanks!

Jim

Lestraveled

Joined May 19, 2014
1,946
Automatic direction finding systems like this are problematic. Here are some things to think about:

- The transmitter and receiver are mounted to dynamically moving mounts, Kayaks move around a lot.
- Even in the best conditions, the receiver will have to adjust for the varying signal strength due to range.
- The transmitter and receiver will get wet and water absorbs IR a lot.

In other words, the blind kayaker could be steered into a rock that is reflecting the IR signal when the correct sensor is blocked by water.

I am not saying it can't be done, but there is a lot to consider.

AnalogKid

Joined Aug 1, 2013
8,251
An IR receiver could be no more complicated that a little hand held unit you wave around till it sees the boat to home in on, then it either lights beeps or shakes.
The receiver would not need anything more complicated than a TV IR receiver looking for a 40KHz signal, then a transistor or two to drive the led or buzzer.
Except that it has to be hands-free and give an indication of relative direction across a 180 degree spread.

ak

Lestraveled

Joined May 19, 2014
1,946
I suggest using a single sensor at the receiver that scans using a mirror. This mirror would be encased in a glass cylinder that spins with the mirror. The advantages would be:
- Water beads would be shed by the centripetal force.
- Any remaining obstructions would be normalized out.
- Detection thresholds are much more effective when you process/average a sweep. In other words, you adjust the gain until there is only one peak.

AnalogKid

Joined Aug 1, 2013
8,251
I thought RF DF cannot resolve an absolute direction. Would not an RF solution leave open the problem of knowing what you are homing in on is upstream or downstream?
Yes, which is why I'm advocating IR. Normally, the inherent directionality of IR emitters and detectors in T-1 and T-1 3/4 packages is a problem to be overcome, in this application it is a natural enhancer of the solution.

ak

Lestraveled

Joined May 19, 2014
1,946
Another thing to consider is false alarm rate.
- It won't work if the blind kayaker is flooded with hits. An automatic gain control (AGC) and adaptive detection will have to resolve the indication to one direction.
- You need to add another indicator to the kayakers vest, an "I don't know" indicator. If the detector can not resolve the correct direction, you have to tell the pilot that.

AnalogKid

Joined Aug 1, 2013
8,251
For concentrating the IR light, I’m wondering if placing each of the LED’s inside of a flashlight reflector cone (eBay /\$1 ea) would help. For that matter, perhaps also the IR receiver modules?
I don't think the emitters need any help. High intensity emitters make their numbers by placing the chip in the domed package such that the dome is a lens, narrowing the emission angle significantly. You mentioned a 30' min. radius, but the first 5 feet are the minimum distance between the emitter at the back end of your boat and the receiver at the front end of his. Easy enough to pick emitters with a radiation angle that fills in the gaps within five feet.

The receivers are the problem, because there we want a narrow angle of view (36 degrees), but most parts (and especially TV remote control parts) are optimized for a wide angle of view. This can be fixed with shutters, baffles, or whatever you want to call physical shields, but it's frustrating to know that the detector chip has sensitivity that is directed away from what we need. I haven't done any digging for narrow-angle receiver parts; might get lucky.

ak

JRiley

Joined Mar 21, 2016
5
Quite true on dynamic environment and IR reflections. I'll likely need to check some average peak value to compare on each poll. Boat rocking, water splashes or other kayakers may block direct path to LED emitters but peak value of reflected radiation should be significantly less than direct path. Job of guide is to try to avoid obstacles getting between. Optimal use would be on flat open water and then determine conditions where this method fails. Backup would be override using direct guidance. Looks like its worth a fun try.
Thanks.

AnalogKid

Joined Aug 1, 2013
8,251
Another thing to consider is false alarm rate.
I'm an all-analog/all-discrete kinda guy, but improving the receiver direction decision means cooking down a few seconds of observations, and nuttin does that like a uC.

Plus, it's a kayak - the one thing it does not do quickly is yaw, so we've got some time to calculate out the effects of roll.

ak

Lestraveled

Joined May 19, 2014
1,946
............ cooking down a few seconds of observations, and nuttin does that like a uC.............
This depends on how fast you are scanning. If you are spinning at 10 revolution per seconds, the threshold can be updated every 100 milli-seconds.

Lestraveled

Joined May 19, 2014
1,946
Another thing, a lot of the IR remote control receivers have the AGC function built in. AGC is the key to making this work and you must be able to control it. In other words, you will have to use a raw detector diode and build your own amplifier with AGC.

Attached is a data sheet for a typical IR "receiver" IC. Notice the block diagram.

Attachments

• 93 KB Views: 3

Lestraveled

Joined May 19, 2014
1,946
Without active control of the gain (AGC), you can not tell the difference between the strong direct signal and a weak (but detectable) signal reflected from a rock.