Parallelized Optical Flow Estimation on GAP8

This project provides the code accompanying the paper entitled Parallelizing Optical Flow Estimation on an Ultra-Low Power RISC-V Cluster for Nano-UAV Navigation. For now the repository contains the most efficient implementation mentioned in the paper (Local Optim. - i.e., avoiding recalculations per point of interest).

Requirements

This project requires the GAP SDK. The demo can either be run in GVSoC or on a physical GAP8 board, e.g., GAPuino or the AI-deck of bitcraze.

Installation

In case you have the GAP SDK not set up, we recommend the usage of the provided docker installation.

• Run docker compose up -d
• Attach to the running container using docker exec -it <container_name> /bin/bash
• In the active terminal execute the following sequence of commands to install the GAP SDK and register for the Autotiler:

cd /home/gap_user/ws/gap_sdk/
GAP_SDK # Select the appropriate GAP8 model
make all # Enter the requested information for the Autotiler when prompted
cd /home/gap_user/ws/src/ # This command brings you to the source folder of this project

Running the Code

The sample project can be executed using the following command

make all run platform=gvsoc

or alternatively on the board (please follow the official guide the connect to a board)

make all run platform=board

The optical flow code will be executed on all eight cluster cores by default. For single core execution change the following line in the main.c file:

uint8_t SINGLE_CORE = 0;

Citing this Work

If you found our work helpful in your research, we would appreciate if you cite it as follows:

@inproceedings{kuhne2022parallelizing,
  title={Parallelizing optical flow estimation on an ultra-low power risc-v cluster for nano-uav navigation},
  author={K{\"u}hne, Jonas and Magno, Michele and Benini, Luca},
  booktitle={2022 IEEE International Symposium on Circuits and Systems (ISCAS)},
  pages={301--305},
  year={2022},
  organization={IEEE}
}