Thank you all for your responses on this matter! I had a follow up question about the M113 after doing some digging around with the documentation for the tracked vehicle and Project Chrono. Looking at the JSON implementation I noticed two interesting features: 1) The amount of links for each side are asymmetrical as one has 63 links and the other has 64, and 2) There is a slight asymmetrical offset in the road wheels as well. I made this discovery after I realized that the M113 simulations I created with the Data Driven Driver consistently veered towards the left instead of following a straight path with full throttle and no steering inputs. It makes sense now as the left track has less track shoes than the right, which would give it that behavior.
Looking at the M113 documentation for reference, I understand that it was also built asymmetrically (at least for the number of individual tracks for each side). Is this the reason why the M113 implementation has similar traits or is there another explanation? I am very curious, especially about the road wheel offset. Thank you all again. Chris On Monday, August 25, 2025 at 7:50:41 PM UTC-4 Radu Serban wrote: > Chris, > > > > I see no reason reinventing the wheel like Harry suggests given that the > PathFollowerDriver already implements these PID controllers (one for > lateral control, the other for longitudinal control) and you can adjust all > parameters (including setting the appropriate gain to 0 to make it a PD > controller). > > > > The reason for the behavior you see is simply because the PID gains must > be properly set for the vehicle and type of maneuvers you attempt to > implement. I estimated the values we have in our demos a long time ago for > a HMMWV-type vehicle. We simply used the same or similar values for other > vehicles, with more or less success. Obviously, the dynamics of the M113 > vehicle are too different from that of a HMMWV for the same set of > controller gains to work similarly well. The demos we provide are for > illustration only. It is implied that a user would need to make > appropriate adjustments (for example in PID controller gains) when changing > the vehicle and/or scenario. > > > > You can get ballpark estimates of the gain values by considering a simple > scalar 2nd order ODE mockup model of the vehicle yaw dynamics. > > > > --Radu > > > > *From:* 'Harry ZHANG' via ProjectChrono <[email protected]> > *Sent:* Sunday, 24 August 2025 18:53 > *To:* ProjectChrono <[email protected]> > *Subject:* [chrono] Re: Steering Controller Question > > > > Hi Chris, > > > > I did similar demo before, controlling M113 tracked vehicle to follow sine > shape path with PID method, the video link is ( > https://uwmadison.box.com/s/j93bhbnkwbdh996bi9k7stk1gisas1g2). Is this > demo what you want to do (to some extend))? In this demo, I implemented a > basic PD controller in the simulation loop given the information of > predefined sine-shape trajectory and vehicle's current state > > > > while I am not sure about the reason why it's not working well with > PathFollowerDriver (might because of the tracked vehicle's skid steering > mechanism is different from the ones one wheeled vehicle), I would > recommend trying to implement a basic PID controller, which you could > easily adjust PID gains given over/under damped scenarios. From experience > (demo I shared above), it would work fine with a customized PID controller. > > > > Best, > > Harry > > > > On Wednesday, August 13, 2025 at 8:31:11 AM UTC-5 [email protected] > wrote: > > Hi all, > > I have been testing some of Project Chrono's features with their vehicle > module and I was curious if anyone can explain this abnormality. > > I created a file that creates a PathFollowerDriver component on the M113 > tracked vehicle to test how the PID controller works. I started with a > straight line path first and was able to get some pretty solid results with > the chosen gains. When I transitioned the same gains to a more complicated > path (the Double Lane Change Path) the previous gains that I had were too > harsh and veered the M113 off of the desired path. > > Comparing that to the wheeled vehicle, no matter the maneuver chosen, the > same gains can be used and the wheeled vehicle would be able to follow the > path from beginning to end with little difficulty. My first thought as to > why this may be happening is because of the time step difference in > simulations (2e-3 in the wheeled vs. 5e-4 in the tracked), but I was > wondering if anyone else may have some thoughts behind what may be going on. > > Thank you, > Chris > > -- > > You received this message because you are subscribed to the Google Groups > "ProjectChrono" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion visit > https://groups.google.com/d/msgid/projectchrono/7600dd11-a7b8-4e50-9120-9b86e46829c8n%40googlegroups.com > > <https://urldefense.com/v3/__https:/groups.google.com/d/msgid/projectchrono/7600dd11-a7b8-4e50-9120-9b86e46829c8n*40googlegroups.com?utm_medium=email&utm_source=footer__;JQ!!Mak6IKo!ISC0Jm4_uECvakm47H_xdTjWI0VJX4rMGT7JxNvkQ-NvNQB-IeG6gyCyBjxJ12UWi4-PIM5nxKknvre4xBhCtYSCmKY$> > . > -- You received this message because you are subscribed to the Google Groups "ProjectChrono" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/projectchrono/863ed9f1-e98b-47a8-a9cd-b977319edc78n%40googlegroups.com.
