VisualServoingforRobots
Vision Servoing of Robots
The main challenge of integrating vision as a sensor in any robotic task is to arrive at image processing times close (or even better, less expensive) to the robot sampling rates. Actually, industrial robotics have control laws that turns around 1000 Hz!!. This is not obviously this is not the case on the kind of applications we have in mind, nevertheless it is necessary to consider that as fast as we can compute the robotic situation we can feedback such an information.
Image Based Visual Servoing
- Feature extraction
- Tracking and control law defined around features
Pose Based Visual Servoing
- Feature extraction
- Pose computation
- Tracking and control law defined around pose
Dealing with performance
In the section ACHIEVING OPTIMAL PERFORMANCE FROM THE C/C++ SOURCE CODE of the book C/C++ Compiler and Library Manual for Blackfin Processors you will find a very nice list of conseils that you have to take into account.
A set of functions that are very useful are those related to counting the time of execution.
start_time = clock();
my_favorite_function();
stop_time = clock();
time_used = ((double) (stop_time - start_time)) / CLOCKS_PER_SEC;
printf(" Execution Time of my_favorite function: %f \n", time_used);