Zane kaber/movement speed #3473

zuperzane · 2025-07-01T04:41:32Z

simplescreenrecorder-2025-07-02_21.39.37.mp4

Description

Makes the goalie dive if it can't make it to the ball on time

Testing Done

Added a tough save to the goalie tactic test

Resolved Issues

I dont want to directly link this issue but helps with #3377

Length Justification and Key Files to Review

Review Checklist

It is the reviewers responsibility to also make sure every item here has been covered

[x ] Function & Class comments: All function definitions (usually in the .h file) should have a javadoc style comment at the start of them. For examples, see the functions defined in thunderbots/software/geom. Similarly, all classes should have an associated Javadoc comment explaining the purpose of the class.
[ x] Remove all commented out code
[ x] Remove extra print statements: for example, those just used for testing
[ x] Resolve all TODO's: All TODO (or similar) statements should either be completed or associated with a github issue

itsarune · 2025-07-02T22:21:57Z

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp

@@ -182,6 +183,48 @@ void GoalieFSM::panic(const Update &event)
    Angle goalie_orientation =
        (event.common.world_ptr->ball().position() - goalie_pos).orientation();

+    Point save_pos               = goalie_pos;


this looks great, if you think that there is more work that could be done, you should add a comment with // TODO(#3377): ... to improve this logic

I'm not sure if I should add much to this but I would like to add similar logic to pass defender the thing is the dribble fsm is the thing doing a good piece of the work in pass defender, should I attempt some predictive logic for the dribble fsm?

I see, then you can skip doing the TODO thing that I suggested. It would be pretty useful to have some similar logic in PassDefender as well (maybe by adding a custom state that uses your iterative algorithm to try to find a good final velocity if the final position isn't reachable in time)

Either way, you should try to abstract out this algorithm so that you can use it in the goalie and the pass defender and put the function in software/ai/evaluation/

Okok I'll combine the prs then?

up to you, I think this one is fairly complete. So we could merge it and then make a new PR for the pass defender later

although if the reviews for this take a long time, you can just merge it into this one

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp

itsarune · 2025-07-02T22:27:51Z

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp

@@ -182,6 +183,48 @@ void GoalieFSM::panic(const Update &event)
    Angle goalie_orientation =
        (event.common.world_ptr->ball().position() - goalie_pos).orientation();

+    Point save_pos               = goalie_pos;


what if we search along the stop_ball_point (the point where the ball intersects the goal segment) rather than the goalie_pos (which right now is the perpendicular intersection point between the goalie to intersecting the ball on the way to the goal line).

My rationale is that if we can't make an intersection in time, the best position will always be along the goalie line because it gives us the maximum time to accelerate before the ball gets there

Hmm am I visualizing this correctly you mean like see if travelling to the ball goal intersection would give you time? Cuz my reasoning is it would be ideal if we had some run off because then we would never need to decelerate till we reach the ball, but if we are travelling towards the goal line we need to stop or else we will hit the inside of the goal, like there would be time spent decelerating essentially, idk how the math works lol

I hope this diagram helps. The intersection point (goalie_pos) is calculated to minimize the distance to the intersection point (dotted line). If we know for sure that we can't arrive at this position in time, then maybe the best position is the intersection along the goal segment. The ball would take longer to arrive at this position, so it gives us the most time to accelerate to a faster velocity to intersect the ball.

The OG plan was the orange destination which due to the overshoot allows the robot to be always accelerating and never have to decelerate, and my thing with the blue line is that it gives the most time but the robot would spend half the time accelerating and half the time decelerating due to not wanting to hit the wall as opposed to somewhere where the robot can freely go past. But I see now that maybe a green line of some kind might be a good mix allowing overshoot and more time. I put a gif as well up top somewhere, it's a weird optimization problem, I'll try and fix it during the pass defender pr. Thanks for the tips 🙏

Okay sounds good 🙏

itsarune

great work! just a few nits from me!

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp

itsarune

LGTM ⚽ 🥅

cleaner code Co-authored-by: itsarune <[email protected]>

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp

williamckha

I tried to field test this today but couldn't get the robots working (unrelated to your PR). Regardless, I think this change looks good! I would second Arun's comment about extracting the interception logic and putting it in a separate func for future reuse

zuperzane · 2025-07-07T03:01:45Z

@itsarune sorry could u recheck it I did the epsilon thing lol

mkhlb · 2025-07-13T00:02:17Z

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp

+        Point new_destination = save_pos;
+        double final_speed    = GOALIE_STEP_SPEED_M_PER_S;
+        bool finished         = false;
+        double max_speed = event.common.robot.robotConstants().robot_max_speed_m_per_s;


What if we are too close to the ball to attain this max speed? We should check that and potentially use another max_allowed_speed_mode to accelerate faster or something (although i doubt our robot can even attain a higher acceleration without slipping) if this speed is insufficient.

If it's too close it won't affect its behaviour but idk it s probably weird code practice, but the function that I check the time to destination uses the maximum possible speed and calculates it for us and mins it against the one that's passed should I repeat that check before passing the soeed? I'm very down to do the acceleration thing though

this could probably be handled in a separate PR. I'm not sure how much more benefit we would get

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp

itsarune

nice work!

zuperzane added 2 commits June 30, 2025 22:17

diving goalie and test

ec3dd45

constant

8d6a82f

zuperzane requested review from itsarune, williamckha and sauravbanna as code owners July 1, 2025 04:41

[pre-commit.ci lite] apply automatic fixes

de97406

itsarune reviewed Jul 2, 2025

View reviewed changes

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp Outdated Show resolved Hide resolved

itsarune reviewed Jul 2, 2025

View reviewed changes

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp Show resolved Hide resolved

zuperzane and others added 5 commits July 2, 2025 21:59

nits

40aab81

merge

bf0c9dc

[pre-commit.ci lite] apply automatic fixes

f1cbec5

flaky test change

349919e

better tests

1cde945

zuperzane requested a review from itsarune July 3, 2025 06:19

itsarune reviewed Jul 4, 2025

View reviewed changes

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp Outdated Show resolved Hide resolved

itsarune previously approved these changes Jul 4, 2025

View reviewed changes

Update src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp

15da7d3

cleaner code Co-authored-by: itsarune <[email protected]>

zuperzane dismissed itsarune’s stale review via 15da7d3 July 4, 2025 06:16

zuperzane commented Jul 4, 2025

View reviewed changes

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp Outdated Show resolved Hide resolved

zuperzane and others added 2 commits July 4, 2025 00:37

Update goalie_fsm.cpp

33fe464

[pre-commit.ci lite] apply automatic fixes

a4b1468

zuperzane requested a review from itsarune July 4, 2025 07:40

williamckha previously approved these changes Jul 4, 2025

View reviewed changes

mkhlb reviewed Jul 13, 2025

View reviewed changes

itsarune reviewed Jul 15, 2025

View reviewed changes

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp Show resolved Hide resolved

itsarune reviewed Jul 15, 2025

View reviewed changes

src/software/ai/hl/stp/tactic/goalie/goalie_fsm.cpp Outdated Show resolved Hide resolved

useless line

b2749b9

zuperzane dismissed williamckha’s stale review via b2749b9 July 17, 2025 17:05

zuperzane requested review from itsarune and williamckha July 17, 2025 18:21

itsarune approved these changes Jul 17, 2025

View reviewed changes

williamckha approved these changes Jul 18, 2025

View reviewed changes

zuperzane merged commit 7e9d289 into UBC-Thunderbots:master Jul 18, 2025
6 checks passed

Zane kaber/movement speed #3473

Zane kaber/movement speed #3473

Uh oh!

Conversation

zuperzane commented Jul 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Testing Done

Resolved Issues

Length Justification and Key Files to Review

Review Checklist

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

itsarune Jul 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zuperzane Jul 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zuperzane Jul 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

itsarune Jul 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

itsarune left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

itsarune left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

williamckha left a comment

Choose a reason for hiding this comment

Uh oh!

zuperzane commented Jul 7, 2025

Uh oh!

mkhlb Jul 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

itsarune left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

zuperzane commented Jul 1, 2025 •

edited

Loading

itsarune Jul 2, 2025 •

edited

Loading

zuperzane Jul 2, 2025 •

edited

Loading

zuperzane Jul 2, 2025 •

edited

Loading

itsarune Jul 4, 2025 •

edited

Loading

mkhlb Jul 13, 2025 •

edited

Loading