Roadmap plan

[vmayoral]

In order to develop a roadmap plan we should review http://dev.ardupilot.com/wiki/building-for-beaglebone-black-on-linux/ and take as much inspiration as we can.

Let's try to write down a roadmap for the first months of 2014 while discussing about it over here.

[vmayoral]

Another interesting approach is the one available over here. Although not following our goal of development of open software/hardware autopilot named BeaglePilot for linux-based computers, we can get inspiration from there as well.

Maybe we can work on a list of requirements and backtrack from there?. These are some ideas:

Goal: development of open software/hardware autopilot named BeaglePilot for
 linux-based computers

Requirements:
   - ROS integrated
   - Well documented
   - Linux-based

I usually like working using the spiral model so i'd recommend to get some working autopilot asap and then build up prototypes from there. This is just the approach i usually take but i'm open to hear other interesting ideas.

My suggestion would be:

• Get a working autopilot on Linux (using the BB or the BBB) and test it with real hardware.
• Document it properly.
• Performance tests.
• Start with ROS integration. ros_control seems a nice approach.

These are just my 2-cents. I'd be happy to discuss it.

[mitul93]

I have some thoughts:

• As Andrew Tridgell suggested xenomai and PRU are good options for low latency response. We should give it a try and check weather it is really helpful or not.
• What's the status of sensors? Drivers are available for all sensors? If not we should start with drivers. Do interfacing with Beaglebone and check sensor outputs.
• ros_control is a good start point. Victor only you can test it right now. We need hardware to test it.

[vmayoral]

Thanks for your comments, we should indeed check Xenomai, RT patch and the PRU.

The guys behind ArduPilot are implementing PWM and PPM at the PRU.

[vmayoral]

I like the idea of making issues out of every task we will focus on. Let's wrap them and divide the work:

- Simple PID implementation and test in hardware.
- PWM generation using PRU
- Realtime evaluation (normal kernel, RT patch, Xenomai)
- open hardware/software ESC development
- ROS integration (ros_control) 

@mitul93 on my board all sensors have drivers already however we can compile a list of tentative sensors that we would like to include and start searching/coding the drivers.

P8 and P9 are free in my boards which means that much stuff can be hooked up ;).

[mitul93]

What are the different sensors? Are we considering replacing/modifying it with better option(s) if we can find a economical solution?

[vmayoral]

Yes, they could be replaced if better options/lower cost is found. The sensors:

• Pressure/Altitude: BMP085 (this one should probably be replaced since it's not manufactured anymore)
• GPS: MTK3339
• Light sensor/range finder: VCNL4000
• IMU: MPU9150

Future integrations:

• Microphone: to be decided but probably a MEMS based
• Camera: Already tested with an UART camera, in the short term Pixy (CMUcam5 is a target since it'll be open hardware)

[vmayoral]

Refer to the README of this repository for an updated roadmap plan that includes our GSOC proposals.