lr-morales
commented
8 years ago Progress report
Check list
- [x] Locate and acquire sensor for testing
- [x] Preliminary system analysis and working test
- [x] Study the output signal
- [ ] System response characterization
- [x] Devise good connection method to RPi2
- [x] Receive signal from sensor
- [x] JdeRobot driver encoding
- [x] Driver testing
Locate and acquire sensor for testing
Searching around the net, finally decided to use one good price sensor system composed by a central unit, a buzzer and a metallic adhesive antenna band.
The antenna of this sensor system is intended to be fixed inside the rear plastic bumper of the car; the central unit and the buzzer are fixed with an adhesive inside the car. The central unit is designed to be connected to 12V - ideally to the reverse light - and a valid ground.
Preliminary system analysis and working test
For a first analysis, soldered a connector to the power input lines to allow a commercial 12V adapter to be used instead a real car; the antenna was partially unrolled on a table leaving a distance of more than 1m to other obstacles in one side.
After powering up the system, the buzzer beeps 3 times to signal that is ready to operate. If there is a bad connection, something wrong with or touching the antenna or another problem, the buzzer beeps with a sequence of deep tones.
When working in good conditions this behaviour can be observed:
- Total insulators (dry wood, cardboard, plastic) don't affect the sensor as it relies in the electromagnetic field detected by the antenna.
- Objects moving away from the sensor are ignored.
- If an object moves excessively near to the sensor (few centimetres), buzzer beeps in similar way as an error. It needs to rearm itself before operating again (rearm time and conditions to be studied).
- If an object moves too fast towards to the sensor, buzzers beeps signalling the excessive speed. Then readjusts the detection zone (need to determine exactly this readjustment).
- If an objects moves in a paced way towards to the sensor, 3 areas are defined (distances to be determined experimentally):
- Far area: buzzer sounds beeps of medium pitch spaced proportionally to the distance to the sensor.
- Near area: buzzer emits a constant high beep.
- Contact area:buzzer emits a constant deep beep.
Study the output signal
In order to connect the sensor to JdeRobot, a method to interface the sensor against some kind of computer is needed. The main output of the system is the buzzer, so a good approach is trying to adapt the signal sent to this buzzer to a device with JdeRobot.
Analysing the signal sent to the buzzer in a quick way with an oscilloscope we get this kind of results:
It seems to be a low frequency PWM-like signal in the range of 1~3V.
After making additional tests, some peaks around 5~7V were detected during power-on phase. In order to avoid receiver input damage, a 7V maximum will be considered unless a significant signal drop is detected during next phases.
System response characterization
TODO
Devise good connection method to RPi2
After some initial designs for adapting a PWM-like output with possible peaks of 7V to the 3.3V maximum GPIO input of a Raspberry Pi 2, an optoisolated aproach was selected.
In the left side of the schematic, Vss is the input signal positive and PGND its reference; the resistor for that side is the current limiter for the optoisolator. In the right side, the 3.3V output, input pin from which the signal will be read and the ground GPIO pins are shown. The resistor for that side is a pull-down resistor (an arbitrary value was used) to ensure that the input pin is set to the reference value when no input is received.
The optoisolator used is an PC817B as it was available for testing at the moment, but any compatible component could be used instead.
In this schematic, when the central unit sends the PWM for the buzzer, the optoisolator input will receive the signal and close de output circuit according to this input.
In order to check the viability of the idea, four steps were taken before making the final connection: 1) Setting the input to a constant 3V and the output to a LED. 2) Setting the input to a constant 3V and the output to a piFace - I/O protected board - connected to a Raspberry Pi B. 3) Setting the input to the central unit and the output to a LED. 4) Connecting the central unit to the piFace board.
Last step was connecting the complete system for verification with ano auxiliary buzzer in order to check the working state.
Receive signal from sensor
For checking the signal received a simple mixed websocket/static webpage server in python 3 was written. This server launch a data acquisition process, a main handler process and a server process. The idea is encapsulate the functionality in these three processes, allowing the modification and adaptation of new modules for each part; specially changing between the piFace and the direct GPIO connection.
In the previous image, one of the capturing tests against the GPIO can be seen. These tests have shown that is possible at first glance to detect quickly enough the PWM signal variations.
JdeRobot driver encoding
Having a working input to a RPi2 GPIO input pin, the next step is to encode a new driver for this sensor. To achieve this, a GPIO controller library, a new dedicated interface and JdeRobot system in the RPi2 are needed.
The driver, called "emSensorDriver", has three explicit threads to do its work:
- A main thread that initializes the driver elements, handles external OS signaling and global elements of the driver.
- An adquisition thread, which captures and queues the selected GPIO signal, running faster tan the other two.
- A signal analysis and logic thread, which extracts information from the captured data and sends information via the ICE interface.
Driver testing
TODO
There are parking electromagnetic sensors in the market that use an anthenna to detect distance to obstacles near the car (under 1m approx). As output they use a buzzer to signal one of three levels on the measured distance. Supporting these sensors opens the door to new applications