-1
\$\begingroup\$

I am looking for a sensor to reliably determine the state of motion of a car (passenger, van, forklift, tractor).

I will not disclose the nature of the project at this time but a similar sensor system could be used to activate a log of travel hours or send a GPS position report on arrival and departure from a destination.

First priority is low power, it must trigger a micro-controller (in deep sleep) only on motion start and stop, my complete system will operate on 2 or 3 coin cells for long periods (not wired to the vehicle, relocatable, cheap).

Second priority is 0% false detection of motion at the cost of response time.

Response time of half minute is optimal, 5 minutes is about the limit.

I require a signal edge on the state change from stationary to in-motion and vice versa and knowledge of which state the car is in.

I have considered trembler switches using springs, ball bearings, conductive liquids (not Hg). Also magnetic field vector sensor to determine rotation of vehicle, might not detect short movements. An accelerometer to sense motion might false on loading activities.

Passive sensors could interrupt from motion. Magnetic and acceleration sensors would need to be polled (every 30 seconds) to determine if changes have occurred. An alternative would be to use a combination with the trembler switch triggering the other sensor to confirm movement and avoid false motion event.

Changes in engine or ignition status, ambient lighting, noise level, occupancy, loading, unloading, construction site proximity must not cause false motion state.

Also intermittent motion in a parking garage, fuel queue, traffic lights or traffic jam should maintain a reliable motion detection.

My current idea is to have a sensitive trembler switch with asymmetrical filtering and long time hysteresis to change.

Are there standard sensors types that are used for this or that would provide me with a simple output I could condition?

\$\endgroup\$
8
  • 2
    \$\begingroup\$ As you have GPS, why not poll the GPS receiver? Also, what do you want to happen when the car is parked on a moving ferry? \$\endgroup\$
    – user16324
    Jul 12, 2015 at 21:19
  • 1
    \$\begingroup\$ Sounds like a serious project. Hope you google things and research the materials not only here. I was involved in a similar project a while back, there was a bunch of sensors, more or less all common sensors you find on market, because no one can detect everything everywhere. \$\endgroup\$
    – user76844
    Jul 12, 2015 at 21:25
  • \$\begingroup\$ @BrianDrummond my project does not have a GPS, not needed and the power consumption would be much too high, to use as a check would be nice except for the underground garage/tunnel problem, just to confirm it would be lovely but much too expensive as I do not need any of the data it provides. I had not thought of ferry (or train) but on reflection it will not matter to me if it has false readings there as long as it recovers when back on the ground. \$\endgroup\$
    – KalleMP
    Jul 12, 2015 at 21:33
  • 1
    \$\begingroup\$ As always... Look, i can't share much, it's not my IP. I can say, that gps is good except parking lots and dense towns like in Europe, so something else is required to at least log events. I think apple patented inertial positioning, which may be part of the solution. \$\endgroup\$
    – user76844
    Jul 13, 2015 at 4:18
  • 1
    \$\begingroup\$ How do you define 'motion'? Could it be the turn of one specified wheel at least some angle? \$\endgroup\$
    – Neil_UK
    Oct 9, 2015 at 9:17

4 Answers 4

3
\$\begingroup\$

As you have hinted in your question, an accelerometer with a 30 second poll interval would probably be the lowest power solution. Tilt switches and piezoelectric sensors aren't going to be sensitive enough to the kind of motion that you are trying to detect.

The uC would power off the accelerometer, sleep for 30s, then wake, power on the accelerometer and take a reading. If there is any motion at that instant then you run a more thorough algorithm for a few seconds before deciding whether to mark the vehicle as moving or not, then go back to sleep.

As for 'standard sensors', probably not as your problem is specific. But a search for low power accelerometers results in the ADXL362 which takes 1.8uA, runs off a wide range of voltages and has SPI so can easily be connected to a uC.

\$\endgroup\$
1
  • \$\begingroup\$ Yes, I am leaning towards having a slow polling system of an accelerometer, (perhaps even a rate gyro or magnetometer for backup), a tilt/vibration sensor might be suitable for waking up from deep sleep if parked overnight. \$\endgroup\$
    – KalleMP
    Oct 19, 2015 at 18:16
1
\$\begingroup\$

LowTech Solution: Place a magnet somewhere on the wheel and have a reed-contact on your board. Use a monoflop with a long time.constant on your board to keep it activated when driving slowly or stopping.

HiTech Solution: Another Idea would be to use the sensor of a wireless optical mouse. They are made to run a long time on a single battery. You "just" have to modify the optics for longer range (10cm - 20cm instead of a few mm).

\$\endgroup\$
1
  • \$\begingroup\$ Not practical. Must work on any vehicle type, car or tractor, new or old. Must not require installation beyond adhesive tape or velcro. \$\endgroup\$
    – KalleMP
    Oct 19, 2015 at 8:07
0
\$\begingroup\$

One solution would be to mount/attach a small motor/generator so that it will be driven by the vehicle's drive shaft (a couple of tie wraps & a rubber drive wheel of the right diameter to provide good motor drive traction, a 2-wire harness to your electronic system, & you've got power!). When the vehicle moves, the motor produces power that eventually switches on a transistor that could be used as a sensor to switch on your electronic system. When the shaft stops turning, you can turn off your circuit in whatever controlled battery-operated way that you desire with a μIC controller.

With sufficient shaft rotation, your motor could provide enough stable power to drive a voltage regulator, charge rechargeable cells, power a GPS, & even send wireless data to a receiver.

Such a system would make a pretty effective real-time tracking system--not that you or anyone else would ever do such a thing. ;)

\$\endgroup\$
2
  • \$\begingroup\$ Not practical. Must work on any vehicle type, car or tractor, new or old. Must not require installation beyond adhesive tape or velcro. \$\endgroup\$
    – KalleMP
    Oct 19, 2015 at 8:05
  • \$\begingroup\$ It may not be practical in all cases, but it certainly would work in some cases. Many solutions are not universal & must be tailored to situational variables. \$\endgroup\$
    – zeffur
    Oct 19, 2015 at 16:24
0
\$\begingroup\$

The easiest would be to tap in the electrical system of the car. You could read wheel speed sensors from the ABS or you could read messages from the vehicle communication bus (mostly CAN). On CAN you have standardized messages that are common for all vehicles. You will get information of the engine parameters (throttle position, rpm, catalytic converter state, ...) as well as from ABS (wheel speed sensors, steering wheel position, ...). With those parameters it is easy and reliable to distinguish in what motion or driving condition the vehicle is.

\$\endgroup\$
1
  • \$\begingroup\$ Not practical. Must work on any vehicle type, car or tractor, new or old. Must not require installation beyond adhesive tape or velcro. \$\endgroup\$
    – KalleMP
    Oct 19, 2015 at 8:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.