Repository navigation
Let Domino's and Fan's oracles run in the default configuration - #200
Merged
Merged
Conversation
Circuit, Laser and Switch looked up their switch joints without physicsClientId, so PyBullet answered from client 0, the first world the process built. Built after any other env (through the RoboDisco wrapper, in a test session, or in scripts/robodisco_getting_started.py) they read the wrong world and failed with "getJointState failed; invalid jointIndex" or "GetJointInfo failed.". The same omission sat in calls that never crash. Laser created its beam bodies in client 0, so the planner's world drew beams into the executing env's world, and its bookkeeping then removed bodies from its own. Coffee joined its cord segments and its plugged-in plug with constraints in client 0, and the Domino fan component read its switch joints from client 0. Each of these calls now names the env's client, as do the few calls that passed it positionally, so tests/test_pybullet_client_ids.py can require physicsClientId= on every PyBullet call that reaches a physics server. Regression tests build Circuit, Laser and Switch after Blocks in one process and check that Laser's beams appear in its own world only. The envs README and the smoke script no longer describe the failure, and the smoke script numbers its seven steps out of seven.
Both oracle approaches crashed on Domino and Fan in the default configuration, for unrelated reasons. Domino's ground-truth processes were written for domino targets (domino_use_domino_blocks_as_target=True, the benchmark setting). By default the targets are hinged flaps and Toppled ranges over them, so the process for a domino falling flat built Toppled(?d1) over a domino and failed LiftedAtom's type check (structs.py:880). That process now adds Toppled only when dominoes are the targets; with hinged targets a fallen domino just stops tilting. No process models a hinged target falling, so both oracles now run and report the goal unreachable. Fan's default scene (the exposed deck since #196) attaches a FanTransferEvaluator to each task. The process planner rewrites the goal BallAtTarget into the grid goal BallAtLoc but kept the evaluator, failing Task's check that an evaluator judges the task's own goal (structs.py:1051). The rewritten task now drops it, as replace_goal_with_alt_goal does; the planner never reads it. oracle_process_planning then runs, but its grid processes cannot reach the goal cell. The oracle approach already ran without crashing. The envs README now lists Domino and Fan among the environments that run but do not solve the test task.
This was referenced Oct 3, 2026
yichao-liang
changed the base branch from
fix-robodisco-cross-env-leak
to
master
October 3, 2026 07:59
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Both oracle approaches crashed on Domino and Fan in the default configuration, for unrelated reasons.
Domino.
The ground-truth processes were written for domino targets (
domino_use_domino_blocks_as_target=True, the benchmark setting).By default the targets are hinged flaps and
Toppledranges over them, so the process for a domino falling flat (DominoTiltingDelete) builtToppled(?d1)over a domino and failedLiftedAtom's type check atstructs.py:880(the type assertion on that line, not the arity check).That process now adds
Toppledonly when dominoes are the targets; with hinged targets a fallen domino just stops tilting.No process models a hinged target falling, so both oracles now run and report the goal unreachable.
Fan.
oraclenever crashed: its NSRTs come from the grid processes, and only the process planner adds the grid, so it reports the physical goal unreachable.oracle_process_planningcrashed for a different reason.The default scene (the exposed deck since #196) attaches a
FanTransferEvaluatorto each task, and the planner rewrites the goalBallAtTargetinto the grid goalBallAtLocbut kept the evaluator, failingTask's check that an evaluator judges the task's own goal (structs.py:1051).The rewritten task now drops the evaluator, as
replace_goal_with_alt_goaldoes; the planner never reads it.The planner then runs, but its grid processes cannot reach the goal cell.
The envs README now lists Domino and Fan among the environments that run but do not solve the test task.
Test plan
--approach oracleand--approach oracle_process_planningonpybullet_dominoandpybullet_fan: all four run to completion (0/1 solved); three of them crashed before.tests/ground_truth_models/test_process_derived_nsrts.py::test_domino_gt_processes_build_in_both_target_modes(both target modes) andtests/envs/test_pybullet_fan_transfer.py::test_grid_task_drops_the_evaluator_of_the_replaced_goal.Stacked on #199.
🤖 Generated with Claude Code