wvmarle:
How would you be able to get distance from a single IR sensor? Signal strength is a very poor indicator of distance. You need something else: a separate ultrasound ping (and measure time of flight difference), or triangulate two IR sensors.
Ultrasound distance sensors reach 3-5 meters so in time of flight mode you can do double this. IR doesn't get much further than that either. In case of a follow-me thing you need to remain closer anyway as both IR and ultrasound are dependent on a clear line of sight, a person walking between the robot and the master will be enough to make them lose contact. The ultrasound may come through but that will be an echo, meaning the distance is off.
Make sure your sensors have a very narrow field of vision, the narrower the more accurately you can home in on the sender's signal. But it's also going to be harder to find the signal in the first place.
So I could basically make an array of sensors: 3 IR sensors for forward, left and right and a single Ultrasound sensor for depth and triangulate the data between all 4 to create a sense of direction and distance from the master.
I understand what your saying about the interference problem with the ultrasonic sensors when having stuff in between. However, the idea of the robot (or the long-term goal) was just to have it follow me pretty close like a pet on a leash. If I speed up my walking, I want it to keep up right behind me. But would the echo delay from the ultrasonic sensor be too great at those distances for acceleration/deceleration of a motor in real-time? I'm thinking like a healthy 1 meter range.
If I have a static range like that, I could also calculate the optimum angle of the sensor to point at my torso. I could even use 2 more IR sensors rigged to a tiny motor (or even a stepper) to control the angle of the Y axis to keep the ultrasonic sensor in focus. The only problem I could foresee with this given the information all of you provided now is taking a sharp left or right turn which would change the gate of my body and lose a lot of surface area from my torso for the ultrasonic sensor to follow. In theory if the IR sensors are fast enough, the robot shouldn't have a problem keeping up. But it sounds like IR's are pretty slow in comparison.
I believe the material I quickly reviewed about those ultrasonic sensors mentioned that they work in nanoseconds. Common sense would tell me that's a lot faster than milliseconds. So I'm not sure if their delay would be the biggest issue in prototyping this.
It's in applications like this where a GPS would seem almost better. But GPS isn't high enough resolution for this close of distances. Then there's wifi. I've worked with enough router's to know that they can actually locate you in proximity to their distance within a house which is why I suggested Bluetooth first. But that idea got shot down pretty fast.
OK final edit for now: it seems like a more low-form factor and efficient way to do this would be to control it with a high resolution camera and some sort of tracking software. But that is way beyond my programming skills or anything I want to tackle right now. That is like some super high level sh**. I just want to get into the electronics for now, heh.
