Blind spot detecor, Ultrasonic vs Radar?

I would like to build a blind spot detector for my motorcycle. I researched a little bit about blind spot detector technology and I narrowed down to two options: 1) use ultrasonic sensors to measure the distance of the vehicles next to the bike or 2) use radar sensors to sense moving vehicles close to the bike.

I am looking for guidance on which of the two sensors can work better with an arduino and if anyone has tried either one before.

With the ultrasonic sensors I found many examples that can measure distance and I think their implementation for a blind spot detector is pretty straight forward. The only concerns I have are that they can not differentiate between a stationary and a moving car. So they would give off a signal next to a parked car.

So I thought of trying a radar sensor. I looked online and there are cheap radar modules in the X-band, the HB100 or HB200A. I have some questions regarding radar sensors.

  1. In the examples I read online these sensors are stationary and they can detect moving objects. Can they work well to detect moving cars if the sensors are mounted on a bike and the bike moves as well?

  2. I also noticed that the above sensors have a wide detection angle of about 70 degrees in the horizontal? Can the angle be reduced to a more narrow range (20 degrees) or is it this strictly a function of the X-band frequency?

  3. The above radar sensors usually have a detection range of 20m. For blind spot detection applications I would want the range to be limited to 5m or 6m otherwise I will get many false alarm from vehicles that are farther away and are not in the blind spot of the bike. Is it possible to limit the maximum detection range? If yes is it as simple as reducing the input voltage to the sensor?

ultrasound will not work very well at all, due to the movement of air.

But there are some high power ultrasonic transducers that may just be up to the job.

Search on the DIYdrones site, as they use them for controllig drone hight when close to the ground and have similar issues with air movement.

Alternatively, you could go for a video camera solution using OpenCV, but it would probably be much more complex and costly and require multiple raspberry Pi boards

The HB100A you cite looks just like the Doppler unit I use in my intrusion detection system. But the key word here is "Doppler": that means it doesn't really sense the presence of an object, just the motion. So if something is moving along in your blind spot at the same speed you are moving -- no signal. I found these units highly effective and false alarm-free indoors, when they are actually tracking motion. But I wouldn't want to trust my safety to one on a moving motorcycle, given the complex environment and inability to pick up zero relative motion.

I think I'd stick with a mirror.

jrdoner: The HB100A you cite looks just like the Doppler unit I use in my intrusion detection system. But the key word here is "Doppler": that means it doesn't really sense the presence of an object, just the motion. So if something is moving along in your blind spot at the same speed you are moving -- no signal. I found these units highly effective and false alarm-free indoors, when they are actually tracking motion. But I wouldn't want to trust my safety to one on a moving motorcycle, given the complex environment and inability to pick up zero relative motion.

I think I'd stick with a mirror.

From your answer I take it that Doppler units measure relative motion not absolute motion. Is that correct?

I didn't know there was a blind spot on a motorcycle. Where is it?

aarg: I didn't know there was a blind spot on a motorcycle. Where is it?

Out the side of the helmet, I suspect. Why you need good mirrors.

Doppler detectors detect movement, but it is very unlikely indeed that something in the blind spot is so closely tracking the motorcycle that there will be no relative movement.


nikosb: The only concerns I have are that they can not differentiate between a stationary and a moving car. So they would give off a signal next to a parked car.

Presumably would. As with all the other sensors you cite.

nikosb: In the examples I read online these sensors are stationary and they can detect moving objects. Can they work well to detect moving cars if the sensors are mounted on a bike and the bike moves as well?

If you adequately constrain their field of view. As per below.

nikosb: I also noticed that the above sensors have a wide detection angle of about 70 degrees in the horizontal? Can the angle be reduced to a more narrow range (20 degrees) or is it this strictly a function of the X-band frequency?

It is a function of the antenna design.

nikosb: Is it possible to limit the maximum detection range? If yes is it as simple as reducing the input voltage to the sensor?

Hmmm. Let me best explain this by an allegory. Can you increase the range of a mobile phone by installing a battery with a higher voltage?

nikosb: 3. The above radar sensors usually have a detection range of 20m. For blind spot detection applications I would want the range to be limited to 5m or 6m otherwise I will get many false alarm from vehicles that are farther away and are not in the blind spot of the bike. Is it possible to limit the maximum detection range? If yes is it as simple as reducing the input voltage to the sensor?

Well, a radar reflection returns a signal strength that is the inverse of the fourth power of the distance, so that should not be a problem! :)

But when this is all done, you will probably get so comfortable with the radar that you will hit a fast moving vehicle because you failed to simply look.

I was commenting on the concept of "reducing the input voltage to the sensor". :roll_eyes:

Paul__B: I was commenting on the concept of "reducing the input voltage to the sensor". :roll_eyes:

I wasn't commenting on your comment.

I saw this doppler unit and it states that you can adjust the detection range from 2m to 16m by turning a potentiometer. I am not sure how the potentiometer works.

In many cars they use radar modules for blind spot detectors. How can they differentiate between a parked car or the road or other stationary objects that are within the range.

nikosb: I saw this doppler unit and it states that you can adjust the detection range from 2m to 16m by turning a potentiometer. I am not sure how the potentiometer works.

It's called a "gain control". It adjusts the amplification of the audio signal.

nikosb: In many cars they use radar modules for blind spot detectors. How can they differentiate between a parked car or the road or other stationary objects that are within the range.

They almost certainly do not (but may have more advanced processing to reject reflections at higher relative speeds).

They are intended for use on motorways which do not have parked cars or other stationary objects, particularly on the driver's side.

In either case both stationary objects (parked cars...) and moving objects have to be distinguished. It's only known that stationary objects move by faster than objects moving in the same heading direction, and slower than objects moving in opposite direction (contraflow).

A doppler sensor will give the speed of some approaching object, relative to the sensor, but nothing about its heading. Similarly a distance sensor can detect the (constant or changing) distance of some other object, but no direction.

All sensors either have a narrow beam, for the detection of objects in a specific direction, or a wide beam that can react on multiple different objects at the same time.

In addition all sensors can be fooled by vibrating surfaces, e.g. when a doppler radar is calibrated by use of a tuning fork.

I have no idea how to build a system for the intended purpose, unless some hard assumptions limit its use to a very specific scenario :-(