animation with the command show(robot) is very, very slow

Within a robotics class, in the control loop, I have the following code to render the robot movement:

show(robot,q_Gen,'PreservePlot',true,'Frames','off');
hold on
plot3(xd(1,1:i),xd(2,1:i),xd(3,1:i),'r')
axis([-.4 .4 -.4 .4 -.1 1])
view(45,35)
drawnow
hold off

The robot has been defined outside of the loop by the command

robot = loadrobot('kinovaGen3','DataFormat','row','Gravity',[0 0 -9.81]);

The animation is very, very slow in any o.s. and PC where it has been tried (Linux, Windows, MAC).

Is it a drawback that can not be avoided?

[Jesse]

Simulink is better for this kind of animation.

[g. a.]

could you please elaborate this?

Answer 1

[Simar]

Hi,

I understand that you are experiencing slow animation speeds across different operating systems and PCs, is not an unavoidable drawback.The performance issue stems from how rendering and updating of the robot's position are handled within the loop. Here are several strategies worth considering to improve the performance:

1. Reduce Rendering Frequency:

If the loop iterates very quickly, consider updating the plot less frequently. For example, only update the robot's position every nth iteration.

2. Optimize Plot Updates:

Instead of redrawing the entire plot in every iteration (which puts load on performance), update only the necessary elements. MATLAB allows you to modify existing plot objects. For instance, update the XData, YData, and ZData properties of the plot object instead of re-plotting it.

3. Use “animatedline” Function:

The “animatedline” function is more efficient for animations where one is appending data points in a loop. It is designed to handle dynamic updates more efficiently than redrawing the plot with “plot3” in each iteration.

4. Limit the Complexity of the Scene:

Ensure that the scene being rendered is not overly complex. Simplifying the scene can significantly improve rendering times. This includes reducing the complexity of the robot model if possible.

5. Profiling and Optimization:

Use MATLAB “profiler” (profile on, profile viewer) to identify the bottlenecks in the code. There might be other parts in the script that are unexpectedly time-consuming.

6. Hardware Acceleration:

Ensure that MATLAB is using the hardware effectively. For instance, MATLAB can leverage GPU for certain operations, which might help in rendering complex scenes more efficiently.

Here is a small adjustment using “animatedline” for the plotting part of the code. This assumes “xd” contains the trajectory points you want to animate.

% Assuming xd is defined 
h = animatedline('Color','r','LineWidth',2); 

axis([-.4 .4 -.4 .4 -.1 1]) 
view(45,35) 

for i = 1:size(xd, 2) 
    % Update robot position here 
    % show(robot, q_Gen(i,:), 'PreservePlot', true, 'Frames', 'off'); 
    
    % Add points to animated line 
    addpoints(h, xd(1,i), xd(2,i), xd(3,i)); 
    
    drawnow limitrate % This helps manage the update rate and improve performance 
end 

By optimizing code and using MATLAB more efficient functions for animations, one should be able to improve the rendering speed of robot's movements.

Please refer to the following documentation links -

• Animatedline- https://www.mathworks.com/help/matlab/ref/animatedline.html?searchHighlight=animated%20line&s_tid=srchtitle_support_results_3_animated%20line
• Profiler- https://www.mathworks.com/help/matlab/ref/profiler-app.html?searchHighlight=profiler&s_tid=srchtitle_support_results_2_profiler
• Profile code to improve performance- https://www.mathworks.com/help/matlab/matlab_prog/profiling-for-improving-performance.html?searchHighlight=profiler&s_tid=srchtitle_support_results_3_profiler
• Addpoints- https://www.mathworks.com/help/matlab/ref/addpoints.html?searchHighlight=addpoints&s_tid=srchtitle_support_results_1_addpoints

Hope this helps!

Best Regards,

Simar

[g. a.]

Thanks for your reply. In my understanding:

1. ok. it works
2. I'm not really sure it is efficient code writing when resorting to an existing library such as the robotic toolbox. In such a case the more convinient is to simply call <show>
3. see 2
4. the scene is empty and the robot model is an existing one. I wouldn't like to enter in the code of an existing library
5. Profiler already gave me the info that it was actually the command <show> the issue
6. I should investigate this

thanks again,

g.

[Umar]

Hi G.A.

To optimize the animation and improve performance, we can make the following modifications to the code:

% Preallocate memory for xd to improve performance
xd = zeros(3, numIterations); % Assuming numIterations is the total number of iterations
% Move the 'show' and 'plot3' functions outside the loop for efficiency
show(robot, q_Gen, 'PreservePlot', true, 'Frames', 'off');
hold on
for i = 1:numIterations
    % Update robot position or configuration
    
    % Update xd values
    xd(:, i) = [x_value; y_value; z_value]; % Update with actual values
    
    % Plot the updated trajectory
    plot3(xd(1, 1:i), xd(2, 1:i), xd(3, 1:i), 'r');
    
    axis([-.4 .4 -.4 .4 -.1 1]);
    view(45, 35);
    drawnow;
end
hold off

By preallocating memory for xd and moving the show and plot3 functions outside the loop, we reduce unnecessary computations and improve the animation's efficiency. This optimization should help in achieving smoother and faster rendering of the robot movement.

Hope this will help resolve your problem.

[Umar]

Sorry for unformatted code

[Walter Roberson]

I think it should be more like

% Preallocate memory for xd to improve performance
xd = zeros(3, numIterations); % Assuming numIterations is the total number of iterations
% Move the 'show' and 'plot3' functions outside the loop for efficiency
show(robot, q_Gen, 'PreservePlot', true, 'Frames', 'off');
hold on
i = 1;
xd(:, i) = [x_value; y_value; z_value]; % Update with actual values
h = plot3(xd(1, 1:i), xd(2, 1:i), xd(3, 1:i), 'r');

axis([-.4 .4 -.4 .4 -.1 1]);
view(45, 35);
drawnow;

for i = 2:numIterations
    % Update robot position or configuration
    
    % Update xd values
    xd(:, i) = [x_value; y_value; z_value]; % Update with actual values
    
    % Plot the updated trajectory
    set(h, 'XData', xd(1, 1:i), 'YData', xd(2, 1:i), 'ZData', xd(3, 1:i));
    drawnow;
end
hold off

[g. a.]

thanks

actually, having the command

showrobot

outside of the loop does not animate the robot... am I missing something?

[Umar]

Hi everyone,

I know it is none of my business to interfere but I could not help notice the show(robot, q_Gen, 'PreservePlot', true, 'Frames', 'off'); function is placed outside the loop but is not animating the robot as expected. So, would it be a better idea to move the show(robot, q_Gen, 'PreservePlot', true, 'Frames', 'off'); function inside the loop. This way, the robot will be animated at each iteration, showing its movement over time.

[Walter Roberson]

We are probably back to the problem of poor performance.