KeiserSose
Nov 10, 2008, 09:30 PM
Hi to everybody,
I am looking to join a team (existing company/start-up/individuals) working on helicopter
UAV project. I was involved in this kind of project for almost 2 years, as a primary
programmer. I have a complete low-level hardware/software setup ready, which includes:
1. IMU board: ARM-7 LPC 2368/87 microcontroller @72MHz with:
- 3 300 deg gyros(Analog Dev)
- 3 5g accelerometers (Analog Dev)
- 3 magnetometers (Honeywell)
All above sensors are read simultaneously at 1.3 kHz rate, down-filtered to 100/50 Hz, 12 bit.
- 5 Hz GPS (ETek) - parsed to [X,Y,Z] position
- 35 Hz altimeter (Intersema) - parsed to the altitude.
- variable rate (0-6m) SRF10 radar - supposed to be used for take-off and landing.
Also optical flow sensor is in mind.
All drivers are in place and produce ready-to-use data for (non-implemented)
Kalman filter. Some calibration algorithms were implemented, but additional
algorithms/equipment must be developed. Some changes on IMU board needed to
fix design problems and implement a few new features. The whole data collection system
takes less than 3% of CPU usage. So, it might be enough for Kalman filter left (at least 18 ms
per solution cycle at 50 Hz).
2. Controller (autopilot) board ARM-7 LPC 2368/87 microcontroller @72MHz with:
- 3 Spectrum (?) 9-channel 2.4 GHz receivers (working with a standard RC) attached to three
115200 UARTs with voting algorithm. Any combination of channels can be used, no mother unit is required.
- 6 channels PWM servos driver with any combination of pass-thru/controlled channels. Any channels mix.
- RPM driver/controller (PID, adaptive PID, neural network were tried).
No special board. MCB2300 is used currently.
Helicopter is controlled by a regular RC in various modes. In the manual mode, controller just passes
ground controls to servos. In fully-automatic mode, RC handles are used as user "pointing" device.
Anything in between is possible.
3. Communication board (CAN to serial gateway) ARM-7 LPC2368/87 microcontroller @72MHz with:
- Aerocomm 2.4 GHz modem (full-duplex communication protocol with GCS over serial connection).
- SD-card recorder
This units records conversations between IMU and controller, records debugging information, transmits data
to the PC-based GCS, supports control over running programs. Optional board. Radio modem also can be
attached to any other board (without SD functionality). However, it is estimated that it will be needed to unload
non-essential tasks from both IMU and controller.
All boards are sitting on 1 Mbit CAN. Original protocol over CAN (light and fast) allows adding more
nodes, as needed. Since it is a three-board design, every module can be re-implemented, or/and replaced,
possibly using another platform, as soon, as a new board supports CAN.
GCS/telemetry recording software under Windows. Allows controlling parameters on any board
in the system. Also records data in binary and csv files and to the database (MS SQL Server). Provides visual
output from sensors and control programs.
All software is original, no 3rd-party drivers and OSes. 90% written in Assembler and highly optimized for
performance. All drivers and whole system are very well tested. I also have tons of software implemented for
ground testing and calibration. There are also some control algorithms (most of them are proven not working
on helicopters) and ideas. Many versions of PIDs were tried.
So, if there is any team with appropriate resources and, preferably, backed with some money, I would be
extremely interested to join. Almost all dirty work is done, so ideas and implementation abilities are needed,
including an airframe(s), pilots, mechanical and electronics production facilities, and so on.
I am in Chicago. I am not a pilot. I am not an electronics/mechanical engineer. I am a programmer (I think
I am a good one) with a strong mathematical background, but I am not a scientists who can actually design
a Kalman filter and/or a flight control algorithm. However, if you can write a system of differential equations,
I am able to understand it and to write a program to solve it. If you can draft a neural network, I can implement
it too.
So, if all of this makes sense to you and you would be interested to work together, shoot me email to:
this_ip_address "at" yahoo "dot" com.
I am looking to join a team (existing company/start-up/individuals) working on helicopter
UAV project. I was involved in this kind of project for almost 2 years, as a primary
programmer. I have a complete low-level hardware/software setup ready, which includes:
1. IMU board: ARM-7 LPC 2368/87 microcontroller @72MHz with:
- 3 300 deg gyros(Analog Dev)
- 3 5g accelerometers (Analog Dev)
- 3 magnetometers (Honeywell)
All above sensors are read simultaneously at 1.3 kHz rate, down-filtered to 100/50 Hz, 12 bit.
- 5 Hz GPS (ETek) - parsed to [X,Y,Z] position
- 35 Hz altimeter (Intersema) - parsed to the altitude.
- variable rate (0-6m) SRF10 radar - supposed to be used for take-off and landing.
Also optical flow sensor is in mind.
All drivers are in place and produce ready-to-use data for (non-implemented)
Kalman filter. Some calibration algorithms were implemented, but additional
algorithms/equipment must be developed. Some changes on IMU board needed to
fix design problems and implement a few new features. The whole data collection system
takes less than 3% of CPU usage. So, it might be enough for Kalman filter left (at least 18 ms
per solution cycle at 50 Hz).
2. Controller (autopilot) board ARM-7 LPC 2368/87 microcontroller @72MHz with:
- 3 Spectrum (?) 9-channel 2.4 GHz receivers (working with a standard RC) attached to three
115200 UARTs with voting algorithm. Any combination of channels can be used, no mother unit is required.
- 6 channels PWM servos driver with any combination of pass-thru/controlled channels. Any channels mix.
- RPM driver/controller (PID, adaptive PID, neural network were tried).
No special board. MCB2300 is used currently.
Helicopter is controlled by a regular RC in various modes. In the manual mode, controller just passes
ground controls to servos. In fully-automatic mode, RC handles are used as user "pointing" device.
Anything in between is possible.
3. Communication board (CAN to serial gateway) ARM-7 LPC2368/87 microcontroller @72MHz with:
- Aerocomm 2.4 GHz modem (full-duplex communication protocol with GCS over serial connection).
- SD-card recorder
This units records conversations between IMU and controller, records debugging information, transmits data
to the PC-based GCS, supports control over running programs. Optional board. Radio modem also can be
attached to any other board (without SD functionality). However, it is estimated that it will be needed to unload
non-essential tasks from both IMU and controller.
All boards are sitting on 1 Mbit CAN. Original protocol over CAN (light and fast) allows adding more
nodes, as needed. Since it is a three-board design, every module can be re-implemented, or/and replaced,
possibly using another platform, as soon, as a new board supports CAN.
GCS/telemetry recording software under Windows. Allows controlling parameters on any board
in the system. Also records data in binary and csv files and to the database (MS SQL Server). Provides visual
output from sensors and control programs.
All software is original, no 3rd-party drivers and OSes. 90% written in Assembler and highly optimized for
performance. All drivers and whole system are very well tested. I also have tons of software implemented for
ground testing and calibration. There are also some control algorithms (most of them are proven not working
on helicopters) and ideas. Many versions of PIDs were tried.
So, if there is any team with appropriate resources and, preferably, backed with some money, I would be
extremely interested to join. Almost all dirty work is done, so ideas and implementation abilities are needed,
including an airframe(s), pilots, mechanical and electronics production facilities, and so on.
I am in Chicago. I am not a pilot. I am not an electronics/mechanical engineer. I am a programmer (I think
I am a good one) with a strong mathematical background, but I am not a scientists who can actually design
a Kalman filter and/or a flight control algorithm. However, if you can write a system of differential equations,
I am able to understand it and to write a program to solve it. If you can draft a neural network, I can implement
it too.
So, if all of this makes sense to you and you would be interested to work together, shoot me email to:
this_ip_address "at" yahoo "dot" com.