Add Potential Field Map Construction sample#39
Conversation
|
@ShisatoYano PTAL |
|
Hi @Pana1v , Thank you for the hard work! The execution example looks great, and it’s clear that the core logic for calculating the attractive and repulsive potentials is working correctly according to the formulas. However, I noticed one point in the execution example that we should address before moving on to the path planning phase: Observation: Currently, it seems the potential field is updated based only on the "instantaneous" sensor readings. As a result, cells representing previously detected obstacles are cleared as soon as they move out of the sensor's current field of view. Why we need a change: To ensure stable path planning and safe navigation, we need an accumulative update approach. If the robot "forgets" an obstacle simply because it isn't looking at it anymore, it might plan a path into a blind spot where a collision could occur. Requested Improvement: Could you update the logic to maintain a persistent Potential Field? Specifically: Memory/Persistence: The field should retain information about previously observed obstacles even when they are outside the current sensor range. Consistency: This persistent field will serve as the foundation for the next step—generating a stable path and velocity profile toward the goal. What do you think about this approach? I believe this will make the navigation much more robust. |
- Add obstacle_memory set to retain previously observed obstacles - Implement incremental update strategy with repulsive cache for efficiency - Add clear_memory() method for resetting the field - Optimize to only recalculate affected cells within influence radius (rho) This ensures stable path planning even when obstacles move outside the sensor's current field of view.
|
@ShisatoYano added persistence |
|
@Pana1v Thanks! GIF file of simulation result has not been updated yet. Please push the change together. |
|
I have updated the gif, PTAL @ShisatoYano |
|
@Pana1v Thanks, but the gif of simulation you updated is terminated at around 6.0 sec before end time 25 sec. Please check and fix this problem. |
|
I have updated the GIF to cover full duration @ShisatoYano |
|
@Pana1v Thank you for quick fixing those problems. I confirmed there is no problem and merged successfully!! |
|
Thank you Shisato! Would definitely love to implement and see how the potential field algorithm would work around dynamic obstacles. |
|
As next step, I will create a new issue to implement a global path planning sample with your potential field map construction. After I created, can I assign the issue to you? |
|
Additionally, I want to announce this PR was merged with a mention to you on X and LinkedIn. Can you tell me your accounts if you have? |
2 participants
new mapping simulation where a vehicle constructs a local potential field map, visualizing attractive forces towards a goal and repulsive forces from obstacles.