Prototype goto controller and test in game #23

New Issue

2026-08-17T16:58:21+02:00

SirStone commented

2026-08-17 16:58:21 +02:00

Blocked by: Goto controller algorithm

Parent map: #18

Question

Does the goto controller feel right when tested in the actual game? Does the tank move fluidly to target positions? Does forward/reverse selection work correctly?

Build a minimal goto controller (whatever algorithm was decided in the goto controller ticket), wire it up with hardcoded goto targets (e.g. cycle between corners), and observe the tank's movement in the Tank Royale UI. Evaluate smoothness, arrival behavior, and reverse-driving logic.

**Blocked by:** [Goto controller algorithm](#20) Parent map: #18 ## Question Does the goto controller feel right when tested in the actual game? Does the tank move fluidly to target positions? Does forward/reverse selection work correctly? Build a minimal goto controller (whatever algorithm was decided in the goto controller ticket), wire it up with hardcoded goto targets (e.g. cycle between corners), and observe the tank's movement in the Tank Royale UI. Evaluate smoothness, arrival behavior, and reverse-driving logic.

SirStone added the wayfinder:prototype label 2026-08-17 16:58:21 +02:00

SirStone commented

2026-08-17 18:48:30 +02:00

Resolution

Goto controller validated in-game via GotoTest prototype bot.

Algorithm

Explicit proportional steering with forward/reverse at 90° bearing threshold, using getNewTargetSpeed from utils.nim for smooth deceleration. No tan(a)/cos(a) trick — explicit branching is more robust near walls.

AimTo controller

directionTo(target) - gunDirection → normalizeRelativeAngle → clamp to ±20°/tick. Gun independent of body via setAdjustGunForBodyTurn(true).

Test results

Diamond circuit (center → north → west → south → east → center) completed twice with smooth arcs
Wall-smash test (waypoint at Y=600, physically impossible): bot reaches Y≈579 (tank radius), arrival threshold 40px catches it, advances correctly
Forward/reverse selection works correctly
Debug graphics (waypoint circles + tracking line) functional
Known edge case: oscillation/stuck after ~1158 ticks on 3rd loop — stuck detector needs longer reversal window. Not a blocker for PPO integration.

Prototype location

GotoTest/ — throwaway bot, single file

## Resolution **Goto controller validated in-game via GotoTest prototype bot.** ### Algorithm Explicit proportional steering with forward/reverse at 90° bearing threshold, using `getNewTargetSpeed` from utils.nim for smooth deceleration. No `tan(a)/cos(a)` trick — explicit branching is more robust near walls. ### AimTo controller `directionTo(target) - gunDirection` → `normalizeRelativeAngle` → clamp to ±20°/tick. Gun independent of body via `setAdjustGunForBodyTurn(true)`. ### Test results - Diamond circuit (center → north → west → south → east → center) completed **twice** with smooth arcs - Wall-smash test (waypoint at Y=600, physically impossible): bot reaches Y≈579 (tank radius), arrival threshold 40px catches it, advances correctly - Forward/reverse selection works correctly - Debug graphics (waypoint circles + tracking line) functional - Known edge case: oscillation/stuck after ~1158 ticks on 3rd loop — stuck detector needs longer reversal window. Not a blocker for PPO integration. ### Prototype location `GotoTest/` — throwaway bot, single file

SirStone closed this issue

2026-08-17 18:48:34 +02:00

SirStone referenced this issue

2026-08-17 18:48:50 +02:00

Command abstraction layer for PPO action space #18

SirStone referenced this issue

2026-08-17 18:51:59 +02:00

PRD: Command abstraction layer for PPO action space #24

SirStone referenced this issue

2026-08-18 16:32:24 +02:00

Command abstraction layer for PPO action space #18

SirStone referenced this issue from a commit