Skip to main content

Development of a new learning algorithm

Over the course of the past 2-3 weeks I decided to take the plunge of learning a new training algorithm which had quite alot of attention in the academic community but was also well constructed for easier implementation. The work by Peng et al on developing the Continuous Forward Algorithm (CFA) was my subject of attention. The overall premise behind implementing this algorithm was to investigate less memory-intensive machine learning algorithm that will not sacrifice accuracy or robustness.


The ability to have a real-time learning system for low-cost electronics (such as the Teensy 3.6) has huge implications for the commercialization and affordable access of intelligent drones specifically for developing and even under-developed economies. The far-reaching impact of having technology leveraging efforts of farmers, herders, game rangers, biologists is the next wave of industrialization. This is the essence of this blog. The journey is very exciting and eventual results even more.

Before I get carried away in dream-walking mode, the next step is to embed this new algorithm into a microcontroller and have it run in parallel through a I2C communication link to the Pixhawk of the quadcopter. This will be foundation for the fault detection and identification mechanism, forming one of the major building blocks in intelligent drones.
Labels:

Comments

  1. Unknown5 February 2018 at 16:40

    I think such innovation would be applicable in biometrics as well, where they do image processing (cleaning) before converting images into templates. Templates currently come in multiple light proprietary formats and bulky ISO interchange formats. I think matching biometric patterns would look different in future.

    ReplyDelete
    Replies
    1. pkantue14 April 2018 at 21:51

      Great comment. The application of such a learning algorithm is far reaching indeed. Let me know if you're active in that research field, I would be keen on doing collaborations.

      Delete

Post a Comment

Popular posts from this blog

Setting up the Tarot T4-3D gimbal on the Pixhawk 2.4.8 with Specktrum dx6 Gen2 toggle switch

So i took the challenge of setting up the Tarot gimbal not just for inherent stable video footage but also the flexibility of controlling it from the radio control. However, I encountered quite a few challenges which made me aware that I'm not the one only in this battle . It's quite clear that the setup of the Tarot gimbal using its own software is completely different from how it's been described in the Ardupilot/Arducopter webpage and in mission Planner. In Mission Planner and it's associated site makes one believe that it should be done through software, only to realize that in actual fact the setup is more complex than that.  After two evenings of trying various combinations, I realized the getting the pixhawk Aux channels to communicate with the T4 gimbal requires the following steps: - The Pixhawk Pin9 (Aux1) needed to be activated to pass through user-chosen channel from the transmitter. For the Dx6 Gen2 it was the channel 6, which can assigned the
2 comments
Read more

Matlab to C/C++ code development - Some learning points

Over the last few years, the engineers at the company have invested both their time and sleepless nights in formulating a process for the development of Machine learning algorithms that will satisfy real-time constraints with minimal RAM usage. This is quite a tall task as per default, that would force one to do their development directly in C language. Although that seems like the right choice, the downside is the direct correlation of the debugging time with algorithm complexity.  Such a time could have been rather used in optimizing the algorithm within the MATLAB environment which has excellent tools for the analysis, plotting and debugging. So it was decided to rather learn the Code generation process with the hope that future algorithm could be designed in a similar fashion without the hassle of the compiler-specific run-time issues. The development of this machine learning algorithm would eventually be implemented in a 32bit, 160Mhz speed, 260KB RAM microcontroller.
1 comment
Read more

The hard climb of innovation

For the last couple of months, our design team has been hard at work at detail development of our drone concept which we hope to make public early 2021. These have been unprecedented times with so many changes within our company: people moving countries, stuck at airports, universities closing and transitioning to online classes and exams; all in the space of one year! Nevertheless, one of the fundamental challenges facing the drone industry in developing countries next year, is how to operate within an environment where shipping of critical parts (amongst other things) has been disrupted due to the covid-19 pandemic. If the most critical items (propellers, batteries, sensors, etc. ) of the system are also associated with the longest lead time, this has a significant impact on the operating cost and service coverage that can be achieved. Unfortunately, there's no easy way of solving this issue except if it was envisioned as part of the development process. But this is seldom the ca
Post a Comment
Read more