ROS wrapper for real-time incremental event-based vision motion estimation by dispersion minimisation (EventEMin). This code was used to demonstrate live incremental motion estimation using a DAVIS346B. We used PlotJuggler to show the motion estimates compared to IMU measurements in a graph similar to the one below:
If you use this code in a scientific publication, please cite the following papers:
- Entropy Minimisation Framework for Event-based Vision Model Estimation
- Robust Event-based Vision Model Estimation by Dispersion Minimisation
- Live Demonstration: Incremental Motion Estimation for Event-based Cameras by Dispersion Minimisation
The authors provide this code in the hope it will be useful for understanding the proposed method, as well as for reproducibility of the results.
For more information and more open-source software please visit the Personal Robotic Lab's website: https://www.imperial.ac.uk/personal-robotics/software/.
This code was tested on Ubuntu 16.04, 18.04 and 20.04, and respective ROS versions, using a DAVIS346B live. It was also tested using bag files recorded by a DAVIS240.
-
EventEMin (no need to install, just ensure all dependencies are met; you do not need to install GSL - GNU): https://github.com/ImperialCollegeLondon/EventEMin
-
rpg_dvs_ros: https://github.com/uzh-rpg/rpg_dvs_ros
We assume you already have a catkin_ws
environment.
On a terminal type:
cd ~/catkin_ws/src
git clone --recurse-submodules https://github.com/ImperialCollegeLondon/event_emin_ros.git
cd ..
catkin build event_emin_ros
Open a terminal and run:
roscore
This package assumes that there are two active publishers for events
and camera info
.
Then, on a terminal type:
rosrun event_emin_ros <model> camera_info:=/subscribe/to/camera_info events:=/subscribe/to/events
For example, assuming you are already running the rpg_dvs_ros
package, to estimate rotational motion type:
rosrun event_emin_ros rotation camera_info:=/dvs/camera_info events:=/dvs/events
The estimated variables are published by default to the topic /event_emin_ros/vars
, and whose type is a standard vector (see message type).
Launch files are also provided for convenience. For example:
roslaunch event_emin_ros global_motion.launch model:=<model> camera_ns:=<camera>
assumes another node is publishing camera_info
and events
messages, either from a live camera of a bag file, on the camera_ns
namespace (by default the camera namespace is /camera
).
The following models are currently supported: affinity
, isometry
, rotation
, similarity
, translation2d
.
The event_model_ros code is licensed under CC BY-NC-SA 4.0. Commercial usage is not permitted.