Autopilot Sensing Devices - CC2541 SensorTag Development Kit

[rgleason]

I've forgotten where I saw this device.
SensorTag simplifies development of sensor apps for smart phones.
TI CC2541 Sensor Tag Development Kit CC2541 SensorTag Development Kit - CC2541DK-SENSOR - TI Tool Folder

SensorTag has six small energy efficient sensors including accelerometer, gyroscope, magnetometer, pressure, temp, etc (see bottom left of this doc)
Quick Start Guide http://www.ti.com/lit/ml/swru324b/swru324b.pdf

The SensorTag is fitted with six sensors and all sensors are chosen to be small, energy efficient and low cost surface mount devices. The sensors use I2C interface and are connected to the same interface bus with separate enable signals. To minimize current consumption all sensors are by default disabled and they are in
sleep mode between measurements. Each sensor can be enabled and read
individually . The The SensorTag is fitted with six sensors and a all sensors are chosen to be small, energy efficient and low cost surface mount devices. The sensors use I2C interface and are connected to the same interface bus with separate enable signals. To minimize current consumption all sensors are by default disabled and they are in sleep mode between measurements. Each sensor can be enabled and read individually.

------

It looks like the sensors might be accurate enough to provide good accelerometer and axis control for steering in seas with several Sensor Tags place at the extremities of the ship (maybe 2 sensor, or maybe 4.

Perhaps using several of this Bluetooth Sensors, with an small program, that runs separately or with Opencpn, we could have better autopilot controls anticipating pitch, yaw, acceleration down a wave, deceleration at the bottom of the wave, turning forces (direction and acceleration) at the bottom, instability forces (direction and acceleration) at the top.

Being unfamiliar with what nmea sentences control an autopilot, I would guess the program might be ablee to output better more informed directions via nmea, than the sensors built into the autopilot.

I think Sean was developing something called CruisePlot (its on his github account) which used arduino sensors, perhaps that would be a possible adaptation for these sensors?

Of course Opencpn, would probably need to develop a more robust bluetooth library to support these CC2541 sensors, as well. I believe the bluetooth does support some android GPS to Bluetooth devices.

All of this is completely outside of the Opencpn scope, but it might be a reasonable addon/plugin.

It seems that technology might be catching up to commercial marine autopilot control solutions. However the actual robust autopilot actuators and devices might still be needed unless a wind/mechanical-electrical self steering rudder with tab is fitted which would reduce the power requirements significantly.

It might even be possible to create a hybrid self-steering device that uses both wind feedback, along with these tag sensors and a very small motor to speed up the response of the self-steering device, for better course holding and response.

[piluso]

You need just only one of these devices to build a autopilot. I designed and constructed my own autopilot around a magnetometer and a accelerometer with a PIC microcontroller with very nice results

The next variable you will need is the rudder position. It's impossible to do a steer control without that variable.

I used an estimator with effect hall sensors that counts the turns of the drive motor to get an approximation of the rudder angle, but in an OpenCpn environment, you could use a NMEA rudder angle indicator for this task; so all the pieces are there.

If you want to improve the response, the gyro is the next sensor you must to use.

To here, all the parts are in the market.

Finally, the power electronics for the motor drive; typically a H bdrige, must to be connected to the PC using some interface for receiving three differents orders: clockwise rotate, counter-clockwise rotate, and stop.

This is where I don't see a pre-built solution. May be, there are some NMEA instructions available for this situation; so a RS-232/485 interface together the power electronics could be the final part of an entire PC based autopilot system.

Regards, David

[rgleason]

I am impressed about your accomplishment. Sean is doing some of the same things.

The next variable you will need is the rudder position. It's impossible to do a steer control without that variable. ...you could use a NMEA rudder angle indicator for this task

Yes, those rudder indicators are pretty expensive.

"If you want to improve the response, the gyro is the next sensor you must to use."

The sensortag has six sensors, gyroscope is one of them. I assume you are talking about a gyrocompass though.

Finally, the power electronics for the motor drive; typically a H bdrige, must to be connected to the PC using some interface for receiving three differents orders: clockwise rotate, counter-clockwise rotate, and stop.

This is where I don't see a pre-built solution. May be, there are some NMEA instructions available for this situation; so a RS-232/485 interface together the power electronics could be the final part of an entire PC based autopilot system.

This is beyond my abilities ...I'd need diagrams and instructions and some programming for the PC.

I think my old ST4000 WheelPilot has 3 wires for control forward, backward, and a neutral. What it could use is better sensors. Have fluxgate but no rudder sensor, and no acceleration sensors that I know of.

[piluso]

The sensortag has six sensors, gyroscope is one of them. I assume you are talking about a gyrocompass though.

I tried to say that the gyroscope, just like the sensortag has, will improve the performance (my autopilot doesn't has gyroscope)
For the power stage an Arduino board could be used attached to an H bridge, there are many projects on the net using that configuration.

I think my old ST4000 WheelPilot has 3 wires for control forward, backward, and a neutral. What it could use is better sensors. Have fluxgate but no rudder sensor, and no acceleration sensors that I know of.

The fluxgate is the traditional technology for autopilots. Magnetometer + accelerometer give you a stabilized compass, it's more accurrate with some degrees of heeling than the fluxgate.
The rudder position it's always needed, an estimation at least. A view to the schematics of the ST4000 shows that uses the back emf technique to estimate the rudder position when there is no rudder angle sensor. Some autopilots have replaced the back emf approach by sensors near the axe of the motor for counting turns and get a better estimated.
Surely you'll have a far better performance with a rudder angle indicator connected to your pilot. It's expensive, but probably that is the sensor you'll need to improve . May be a local supplier could lend you one for testing.
Looking at the schematics again, the rudder position is just a (good quality) potentiometer, in the range of 5k-10k; red to one end, blue to middle and green to the other end, no warranties

[tonyappuk]

My proposed autopilot uses an Arduino to interface from the PC to the wind vane input to a very old Autohelm (starboard on, port on and off). This includes the electronics and motor to operate the tiller. Not having a tiller position sensor I intend to generate a notional tiller position by measuring the time the tiller motor operates. On my boat it takes 5 secs to go from centred to hard over. Any errors will be corrected when the next cycle of GPS calculations are made. At least that's the theory. But I'm only doing this as a hobby for guiding my boat in the Solent not a transatlantic voyage in a supertanker!
Tony

[rgleason]

Piluso [David]

Thank you.
Looks like a rudder sensor is definitely in order, maybe one is available second hand or with a homemade potentiometer which is actually quite easy to do, but then there are electronics parts to just connect to seatalk...

Found these links which might be useful... electronics arudino secondary boards and software and drivers with some directions. Have not looked at it carefully yet. Do you think this system would provide better control than my 1993 ST4000 Raytheon Autohelm Wheelpilot (assuming with rudder indicator) has?

Its released as an open source project (GPL). Its in an early but useable stage, and developing fast. Code is on ,
https://github.com/rob42/freeboard-server
See the software interface here
Freeboard [MeshCMS]

Wind Sensor
Anemometer/Wind Vane, PRO
Replacement part costs are <1/4 of Raymarine!

Arduino Board - with Seatalk or NMEA (chose by setting jumpers)
Freeboard Interface Board v1.2 [MeshCMS]
Orientation Solution
ArduIMU V3 features the new MPU-6000 that includes 3 axis gyros & accells built-in, the latest 3 axis I2C magnetometer HMC-5883L and the classic but very robust Arduino Atmega328 running at 16Mhz...
Connection to Arduino by shield and ribbon cable.

MPU-6000 uses SPI for max performance.
Specifications:

• Tri-Axis angular rate sensor (gyro) with a sensitivity up to 131 LSBs/dps and a full-scale range of ±250, ±500, ±1000, and ±2000dps
• Tri-Axis accelerometer with a programmable full scale range of ±2g, ±4g, ±8g and ±16g
• Reduced settling effects and sensor drift by elimination of board-level cross-axis alignment errors between accelerometers and gyroscopes
• Digital Motion Processing™ (DMP™) engine offloads complex MotionFusion, sensor timing synchronization and gesture detection with supported software (not yet currently supported in DIY Drones code)

Specs

• 1 x 5V TTL serial for GPS (EM406GPS just plugs on, but others can be used.)
• 1 x Seatalk or NMEA (chose by setting jumpers)
• 2 x NMEA
• 1 x NMEA (talker only)
• 1 x 5V TTL Serial (to drive autopilot controller)
• 2 x 3A MOSFETs, 1 to drive autopilot engagement solenoid, and 1 to power external alarms
• 1 x log, 5V, for paddlewheel sensor
• 1 x wind - currently supports Peet Bros anemometer
• 3 x upper/lower level alarmed analogue sensor inputs
• 1 x CAN Bus (NMEA2000), or optionally 2 more simple on/off 5V sensors
• 1 x ethernet - so you can connect to the RaspberryPi/Beaglebone/UDOO without opening your box.

Not sure of the connections here, but it looks like it has 1 GPS, 1 Orientation, 1 Seatalk/Nmea, 2 Nmea, 1 Nmea out, 1+ 2 Autopilot drive, 1 log 5v for paddlewheel sensor (is this the speedo?), 3 upper/lower alarm, 1 canbus, 1 ethernet

It looks like you could control the autopilot with software this way. Don't know how the Wheelpilot control head would be involved.

[rgleason]

There's a newer version being developed which also has altitude and analog temp.
https://code.google.com/p/ardu-imu/wiki/ArduIMUV4

Like its predecessor, ArduIMU V4 is an Inertial Measure Unit plus an Arduino-compatible processor that can run on an Attitude Heading Reference System (AHRS) code based on Mahony’s DCM filter algorithm with Madgwick’s magnetic distortion compensation. In addition to the 3-axis gyro sensors, accelerometers & magnetometers and GPS port, the hardware now includes a dynamic analog temperature sensor, ambient light sensor and pressure sensor. T

[rgleason]

Overall summary Technical [MeshCMS]
Pololu High-Power Motor Drivers
Pololu - Motors and Gearboxes

[piluso]

Tony:

Not having a tiller position sensor I intend to generate a notional tiller position by measuring the time the tiller motor operates

I used the same approach in the firsts intents, the problem was that the timing, a little different travel time for motor forward/ motor back. So there was a cumulative error and after 2 or 3 minutes the controller calculates the rudder is at center position; while in fact was many degrees to any side. You can give it a try, but think about the possibility of need to sensing or estimating the rudder position in some other way.

rgleason:

Do you think this system would provide better control than my 1993 ST4000 Raytheon Autohelm Wheelpilot (assuming with rudder indicator) has?

I don't know this system, so I can't say which is better. But for sure, your 1993 ST4000 is a serious steering system, and with the rudder feedback must work fine.

It looks like you could control the autopilot with software this way. Don't know how the Wheelpilot control head would be involved.

The WeelPilot will do all the work

A Wind Vane, GPS, Fluxgate (or an IMU) are references for the pilot, you select one of them and the controller calculates the "amount" of steering to correct the error. The "amount" of steering is the desired angle of the rudder for the correction, so we see again the importance of the rudder position.

I'm not 100% sure, but using OpenCPN, you can take all the NMEA sources and send it to the Autohelm where you can select the source for steering reference. The fluxgate option is always there for standalone operation.