jalf/TIR_PROJ
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Checkpoint 6

Browse Source
This commit is contained in:
Johnny Fernandes
2026-05-11 10:35:48 +01:00
parent b457155538
commit fce0e0c786
27 changed files with 194 additions and 704 deletions
Download Patch File Download Diff File Expand all files Collapse all files
controllers/shepherd_dog/shepherd_dog.py
+83 -269
View File
@@ -4,52 +4,39 @@ Mode is selected by ``HERDING_MODE`` (env var, or via the
``herding_runtime.cfg`` file the launcher writes since Webots strips
env vars on some setups):
strombom → canonical Strömbom collect/drive heuristic.
sequential → single-target "pin and push" — drives the sheep
closest to the pen.
bc → behaviour-cloned MLP, trained on Strömbom demos via
sim. Default policy directory: training/runs/bc.
rl → KL-regularised PPO fine-tune of the BC policy. Same
obs/action space as bc; refines time-to-pen via
environment reward while staying anchored to bc.
Default policy directory: training/runs/rl.
dagger → DAgger data collection. Reads sheep ground-truth
via the receiver, computes the active-scan teacher's
recommended action at every step, drives with either
the teacher (HERDING_DAGGER_DRIVER=teacher, default)
or the loaded student (=student), and logs each
(lidar_stacked_obs, teacher_action) pair. On exit
dumps to ``training/dagger/dagger_<ts>.npz`` for
``tools.dagger_merge_train`` to consume.
strombom → canonical Strömbom (2014) collect/drive heuristic
wrapped in ActiveScanTeacher (opening rotation +
walk-to-centre when the tracker briefly empties).
sequential → single-target "pin-and-push", same wrapper.
bc → behaviour-cloned MLP, trained on Strömbom demos.
Default policy: training/runs/bc/policy.zip.
rl → KL-regularised PPO fine-tune of bc. Same obs/action
space as bc; refines time-to-pen via reward while
staying anchored to bc.
Default policy: training/runs/rl/policy.zip.
Sheep perception
----------------
The dog now perceives sheep through its **front-mounted 140° LiDAR**
(``protos/ShepherdDog.proto``: 180 rays, 12 m max range). Each step
the controller:
The dog perceives sheep through its **front-mounted 140° LiDAR**
(``protos/ShepherdDog.proto``: 180 rays, 12 m max range). Each step:
1. Reads ``lidar.getRangeImage()``.
2. Runs ``herding.lidar_perception.detections_from_scan`` to cluster
returns into world-frame ``(x, y)`` sheep estimates.
3. Folds those into a ``herding.sheep_tracker.SheepTracker`` which
maintains last-seen positions for sheep currently out of the
FOV and latches "penned" once a track disappears near the gate.
2. Runs ``herding.perception.lidar_perception.detections_from_scan``
to cluster returns into world-frame ``(x, y)`` sheep estimates.
3. Folds those into a ``SheepTracker`` which maintains last-seen
positions for sheep currently out of FOV and latches "penned"
once a track crosses the gate plane south.
The output of step 3 is a ``{name: (x, y)}`` dict shaped exactly like
the receiver-based one we used to consume — so Strömbom, Sequential
and the BC obs builder run unchanged. The sheep→dog Emitter/Receiver
link is still up (kept passively for compatibility) but its messages
are *not* used for control.
Sheep ``emitter`` messages are read **for diagnostic logging only**
(GT_penned counter + auto-finish sentinel); they are never used to
drive the policy. Perception for control comes entirely from LiDAR.
All modes share the same low-level differential-drive controller
(``herding.diffdrive.velocity_to_wheels`` with cos(err)-clamped forward
speed), so switching modes does not retune actuation.
A safety supervisor enforces the "dog stays out of the pen" invariant:
if the action would push the dog past ``DOG_SOUTH_LIMIT`` it is
overridden with a north-driving correction. RL fallback: if the policy
zip can't be loaded (SB3 missing, file missing), the controller drops
to strombom mode automatically.
Auto-finish
-----------
When the dog observes (via GT, read off the receiver) that all sheep
are penned, it writes ``training/.run_done`` and the launcher
(``tools/run_webots.sh``) detects it and closes Webots. This keeps
batch evaluation runs bounded.
"""
import math
@@ -62,26 +49,27 @@ _PROJECT_ROOT = os.path.normpath(os.path.join(_HERE, "..", ".."))
if _PROJECT_ROOT not in sys.path:
sys.path.insert(0, _PROJECT_ROOT)
import numpy as np
from controller import Robot
from herding.active_scan import ActiveScanTeacher
from herding.control import modulate_speed_near_sheep
from herding.diffdrive import velocity_to_wheels
from herding.geometry import (
from herding.control.active_scan import ActiveScanTeacher
from herding.control.modulation import modulate_speed_near_sheep
from herding.control.sequential import compute_action as sequential_action
from herding.control.strombom import compute_action as strombom_action
from herding.obs import build_obs
from herding.perception.lidar_perception import detections_from_scan
from herding.perception.sheep_tracker import SheepTracker
from herding.world.diffdrive import velocity_to_wheels
from herding.world.geometry import (
DOG_MAX_LINEAR, DOG_MAX_WHEEL_OMEGA,
DOG_SOUTH_LIMIT, DOG_WHEEL_RADIUS,
PEN_ENTRY, is_penned_position,
)
from herding.lidar_perception import detections_from_scan
from herding.obs import OBS_DIM, build_obs
from herding.sequential import compute_action_debug as sequential_action_debug
from herding.sheep_tracker import SheepTracker
from herding.strombom import compute_action as strombom_action
from herding.strombom import compute_action_debug as strombom_action_debug
# ---------------------------------------------------------------------------
# Mode selection
# Mode + policy resolution
# ---------------------------------------------------------------------------
def _load_runtime_config():
@@ -122,8 +110,8 @@ def _resolve_policy_dir(mode: str) -> str:
1. HERDING_POLICY_DIR env var or runtime-cfg entry, if it points
to a real directory.
2. Mode-specific default:
bc → training/runs/bc (Strömbom-imitated MLP)
rl → training/runs/rl (KL-PPO fine-tune of bc)
bc → training/runs/bc (Strömbom-imitated MLP)
rl → training/runs/rl (KL-PPO fine-tune of bc)
3. Fall back to bc.
All checkpoints are frame-stacked K = 4; ``policy_loader`` reads
the stacking factor from the policy's observation space.
@@ -135,60 +123,41 @@ def _resolve_policy_dir(mode: str) -> str:
mode_default = {
"bc": os.path.join(_PROJECT_ROOT, "training", "runs", "bc"),
"rl": os.path.join(_PROJECT_ROOT, "training", "runs", "rl"),
"dagger": os.path.join(_PROJECT_ROOT, "training", "runs", "bc"),
}
primary = mode_default.get(mode, mode_default["bc"])
if os.path.isdir(primary):
return primary
# Fall back to BC if the requested checkpoint isn't there yet
# (e.g., user asked for `rl` before training the fine-tune).
fallback = mode_default["bc"]
if os.path.isdir(fallback):
return fallback
return env_dir or primary
_VALID_MODES = ("bc", "rl", "strombom", "sequential", "dagger", "diag")
# Back-compat: an old config saying HERDING_MODE=rl meant "the BC policy".
# We now use `rl` strictly for the KL-PPO fine-tune. If the rl
# directory isn't present, _resolve_policy_dir below silently falls
# back to bc, preserving the old behaviour.
_VALID_MODES = ("bc", "rl", "strombom", "sequential")
if MODE not in _VALID_MODES:
print(f"[dog] unknown HERDING_MODE={MODE!r}; defaulting to strombom.")
MODE = "strombom"
DAGGER_DRIVER = (os.environ.get("HERDING_DAGGER_DRIVER")
or _runtime_cfg.get("HERDING_DAGGER_DRIVER")
or "teacher").lower()
if DAGGER_DRIVER not in ("teacher", "student"):
DAGGER_DRIVER = "teacher"
POLICY_DIR = _resolve_policy_dir(MODE)
policy_handle = None
if MODE in ("bc", "rl", "dagger"):
if MODE in ("bc", "rl"):
print(f"[dog] resolved POLICY_DIR={POLICY_DIR} exists={os.path.isdir(POLICY_DIR)}")
try:
from policy_loader import load as _load_policy
policy_handle = _load_policy(POLICY_DIR)
print(f"[dog] policy loaded from {POLICY_DIR}")
except Exception as exc:
if MODE in ("bc", "rl"):
print(f"[dog] policy load failed ({exc!r}); falling back to strombom.")
MODE = "strombom"
else:
# In dagger mode, no policy is fine if driver=teacher.
print(f"[dog] policy load failed ({exc!r}); dagger driver forced to teacher.")
policy_handle = None
print(f"[dog] running in mode={MODE}"
+ (f" driver={DAGGER_DRIVER}" if MODE == "dagger" else ""))
print(f"[dog] policy load failed ({exc!r}); falling back to strombom.")
MODE = "strombom"
print(f"[dog] running in mode={MODE}")
# ---------------------------------------------------------------------------
# Action smoothing + safety supervisor
# Control parameters
# ---------------------------------------------------------------------------
ACTION_SMOOTH = 0.55 # was 0.35; bumped for less frame-to-frame action jitter
prev_action = (0.0, 0.0)
ACTION_SMOOTH = 0.55 # EMA on (vx, vy) — kills frame-to-frame jitter
RUN_DONE_FILE = os.path.join(_PROJECT_ROOT, "training", ".run_done")
def safety_clamp(vx: float, vy: float, dog_x: float, dog_y: float) -> tuple:
@@ -202,10 +171,6 @@ def safety_clamp(vx: float, vy: float, dog_x: float, dog_y: float) -> tuple:
return (vx, vy)
# ---------------------------------------------------------------------------
# Driving
# ---------------------------------------------------------------------------
def drive(vx: float, vy: float, left_motor, right_motor, compass, motor_max: float):
if math.hypot(vx, vy) < 1e-3:
left_motor.setVelocity(0.0)
@@ -245,12 +210,9 @@ receiver = robot.getDevice("receiver"); receiver.enable(timestep)
emitter = robot.getDevice("emitter")
lidar = robot.getDevice("lidar"); lidar.enable(timestep)
# The receiver channel from sheep is no longer consumed for perception
# (kept enabled in case any peripheral tooling reads it). Sheep
# positions come exclusively from the LiDAR + tracker pipeline below.
tracker = SheepTracker()
# Cosmetic ear motors — ignored by control logic but keep them animated.
# Cosmetic ear motors — animated; not used by control.
left_ear = robot.getDevice("left ear motor")
right_ear = robot.getDevice("right ear motor")
left_ear.setPosition(float("inf"))
@@ -266,75 +228,26 @@ EAR_RATE = 8.0
# Main loop
# ---------------------------------------------------------------------------
# Active sheep positions come from the LiDAR-fed tracker each step;
# penned_set is the tracker's ``get_penned_set()`` call. We drain the
# receiver queue without consuming it, so the small backlog of sheep
# pings can't grow unbounded.
step_count = 0
# Analytic-teacher wrapper (instantiated lazily so RL/BC modes don't pay
# the import-time cost). Each gets the same ActiveScanTeacher treatment:
# rotate-on-empty, walk-to-centre, near-sheep speed modulation.
analytic_teacher = None
if MODE in ("strombom", "sequential"):
base_fn = strombom_action if MODE == "strombom" else sequential_action
analytic_teacher = ActiveScanTeacher(base_fn)
import atexit
import time
import numpy as _np
# DAgger state ----------------------------------------------------------
# Logged each step in dagger mode: (stacked_lidar_obs, teacher_action).
DAGGER_LOG_OBS: list = []
DAGGER_LOG_ACT: list = []
# Diagnostic mode buffer (one dict per step).
DIAG_BUF: list = []
# Frame stack buffer the controller maintains itself when dagger mode is
# active — the stacked obs we log must match what the policy sees so the
# downstream BC consumes (stacked_obs, teacher_action) pairs cleanly.
_FRAME_STACK = (policy_handle.frame_stack if policy_handle is not None else 4)
_dagger_buffer: list = []
# Active-scan teacher operates on GT (read from receiver).
_dagger_teacher = ActiveScanTeacher(strombom_action) if MODE == "dagger" else None
# GT positions accumulated from the receiver (sheep emit their xy each step).
# GT positions from sheep emitters — used **only** for the auto-finish
# sentinel and the GT_penned diagnostic line. Never fed into control.
_gt_sheep: dict = {}
_run_done = False
_DAGGER_RUN_TS = int(time.time()) # one file per controller run
_DAGGER_DUMPED = False
# Sentinel that the auto-collection script polls — empty file written
# when this controller decides the run is "done" (all sheep penned, by
# GT). The launcher then kills Webots and moves on without waiting out
# its timeout. Honoured only in dagger mode.
_DAGGER_DONE_FILE = os.path.join(_PROJECT_ROOT, "training", "dagger", ".DONE")
def _dump_dagger_log():
"""Save accumulated (obs, teacher_action) pairs to disk on exit.
Webots may SIGKILL the controller, so the loop also calls this every
DAGGER_FLUSH_STEPS so we lose at most a few seconds of data per run.
Idempotent — repeated calls overwrite the same file with the latest
accumulated buffer.
"""
global _DAGGER_DUMPED
if MODE != "dagger" or not DAGGER_LOG_OBS:
return
out_dir = os.path.join(_PROJECT_ROOT, "training", "dagger")
os.makedirs(out_dir, exist_ok=True)
out_path = os.path.join(out_dir, f"dagger_{_DAGGER_RUN_TS}.npz")
obs_arr = _np.stack(DAGGER_LOG_OBS).astype(_np.float32)
act_arr = _np.stack(DAGGER_LOG_ACT).astype(_np.float32)
_np.savez(out_path, obs=obs_arr, actions=act_arr)
if not _DAGGER_DUMPED:
print(f"[dog dagger] wrote {len(DAGGER_LOG_OBS)} pairs → {out_path}")
_DAGGER_DUMPED = True
DAGGER_FLUSH_STEPS = 500
atexit.register(_dump_dagger_log)
prev_action = (0.0, 0.0)
step_count = 0
while robot.step(timestep) != -1:
step_count += 1
# Drain receiver. In every mode we capture GT for the diagnostic
# log line — perception still comes from LiDAR, the GT is read-only.
# Drain sheep emitter messages → GT (diagnostic only).
while receiver.getQueueLength() > 0:
msg = receiver.getString()
receiver.nextPacket()
@@ -350,115 +263,28 @@ while robot.step(timestep) != -1:
n = compass.getValues()
dog_heading = math.atan2(n[0], n[1])
# ---- LiDAR perception → tracker → sheep_positions dict ----
ranges = _np.asarray(lidar.getRangeImage(), dtype=_np.float32)
# ---- LiDAR perception → tracker → active sheep positions ----
ranges = np.asarray(lidar.getRangeImage(), dtype=np.float32)
detections = detections_from_scan(ranges, dog_xy[0], dog_xy[1], dog_heading)
sheep_positions = tracker.update(detections)
penned_set = tracker.get_penned_set()
# ---- Diagnostic mode: dump the first DIAG_STEPS scans + GT to disk.
if MODE == "diag":
DIAG_STEPS = 80
if step_count <= DIAG_STEPS:
DIAG_BUF.append(dict(
step=step_count,
ranges=ranges.copy(),
dog_x=dog_xy[0], dog_y=dog_xy[1], dog_h=dog_heading,
gt_sheep=dict(_gt_sheep),
detections=list(detections),
))
if step_count == DIAG_STEPS:
_diag_path = os.path.join(_PROJECT_ROOT, "training", "dagger",
f"diag_{int(time.time())}.npz")
os.makedirs(os.path.dirname(_diag_path), exist_ok=True)
_np.savez(
_diag_path,
ranges=_np.stack([d["ranges"] for d in DIAG_BUF]),
dog_xy=_np.array([[d["dog_x"], d["dog_y"]] for d in DIAG_BUF],
dtype=_np.float32),
dog_h=_np.array([d["dog_h"] for d in DIAG_BUF], dtype=_np.float32),
# Per-step GT serialised: max-pad to 10 sheep.
gt_xy=_np.array([
[list(d["gt_sheep"].get(f"sheep{i}", (1e9, 1e9)))
for i in range(1, 11)]
for d in DIAG_BUF
], dtype=_np.float32),
detections=_np.array([
len(d["detections"]) for d in DIAG_BUF
], dtype=_np.int32),
)
print(f"[dog diag] wrote {DIAG_STEPS} scans → {_diag_path}")
# Build the single-frame LiDAR obs (matches what the env produces).
sheep_xy_list = list(sheep_positions.values())
sheep_penned_list = [False] * len(sheep_xy_list)
single_obs = build_obs(dog_xy, dog_heading, sheep_xy_list, sheep_penned_list)
# Maintain our own frame stack so logged obs == what policy sees.
if not _dagger_buffer:
_dagger_buffer = [single_obs.copy() for _ in range(_FRAME_STACK)]
else:
_dagger_buffer.append(single_obs)
if len(_dagger_buffer) > _FRAME_STACK:
_dagger_buffer = _dagger_buffer[-_FRAME_STACK:]
stacked_obs = _np.concatenate(_dagger_buffer, axis=0).astype(_np.float32)
# ---- Action selection ----
if MODE == "diag":
# Diagnostic mode: rotate in place so the captured scans cover
# all 360° of view from one position. Target = heading + π →
# cos(err) clamps forward to ~0, the dog spins.
_t = dog_heading + math.pi
vx, vy = math.cos(_t), math.sin(_t)
elif MODE == "dagger":
# Teacher: active-scan + Strömbom on GT (active sheep only).
gt_active = {name: xy for name, xy in _gt_sheep.items()
if not is_penned_position(xy[0], xy[1])}
t_vx, t_vy, _mode_str = _dagger_teacher(
dog_xy, dog_heading, gt_active, PEN_ENTRY,
)
# Student (if a policy is loaded).
s_vx, s_vy = None, None
if policy_handle is not None:
action = policy_handle.predict(stacked_obs)
s_vx, s_vy = float(action[0]), float(action[1])
# Drive selection.
if DAGGER_DRIVER == "student" and policy_handle is not None:
vx, vy = s_vx, s_vy
else:
vx, vy = t_vx, t_vy
# Always log the teacher action (this is the supervision signal).
DAGGER_LOG_OBS.append(stacked_obs.copy())
DAGGER_LOG_ACT.append(_np.array([t_vx, t_vy], dtype=_np.float32))
elif MODE in ("bc", "rl") and policy_handle is not None:
# Pass the single-frame obs; the policy_loader maintains its own
# frame stack internally. Both bc and rl use the same control
# interface — the only difference is which checkpoint loaded.
if MODE in ("bc", "rl") and policy_handle is not None:
action = policy_handle.predict(single_obs)
vx, vy = float(action[0]), float(action[1])
elif MODE in ("strombom", "sequential"):
# Wrap the analytic teacher in ActiveScanTeacher so the dog
# rotates / walks-to-centre when the tracker briefly empties,
# instead of going idle. Without this wrapper, the first 2 s
# of LiDAR-blind operation kills the run because Strömbom and
# Sequential both return (0, 0) when there are no positions.
if "_analytic_teacher" not in globals():
from herding.sequential import compute_action as sequential_action
_analytic_teacher = ActiveScanTeacher(
strombom_action if MODE == "strombom" else sequential_action
)
vx, vy, _mode_str = _analytic_teacher(
else:
vx, vy, _mode_str = analytic_teacher(
dog_xy, dog_heading, sheep_positions, PEN_ENTRY,
)
# Shared post-process: speed modulation near sheep. Applies to bc,
# rl, strombom, sequential — every mode where the action source is
# nominally unit-magnitude. In dagger mode the active-scan teacher
# has already modulated, and the diag mode action is hand-built for
# rotation; both skip.
if MODE not in ("dagger", "diag"):
vx, vy = modulate_speed_near_sheep(vx, vy, dog_xy, sheep_positions)
# Near-sheep speed modulation (shared by every mode).
vx, vy = modulate_speed_near_sheep(vx, vy, dog_xy, sheep_positions)
# EMA smoothing — reduces oscillation from policy or Strömbom flips.
# EMA smoothing — kills frame-to-frame action jitter.
vx = ACTION_SMOOTH * prev_action[0] + (1.0 - ACTION_SMOOTH) * vx
vy = ACTION_SMOOTH * prev_action[1] + (1.0 - ACTION_SMOOTH) * vy
@@ -469,7 +295,7 @@ while robot.step(timestep) != -1:
drive(vx, vy, left_motor, right_motor, compass, MOTOR_MAX)
emitter.send(f"dog:{dog_xy[0]:.4f}:{dog_xy[1]:.4f}")
# Cosmetic ear wiggle — purely visual.
# Cosmetic ear wiggle.
ear_phase += 0.12
ear_pos = EAR_AMPLITUDE * math.sin(ear_phase)
left_ear.setVelocity(EAR_RATE)
@@ -477,38 +303,26 @@ while robot.step(timestep) != -1:
left_ear.setPosition(ear_pos)
right_ear.setPosition(-ear_pos)
# --- Early-stop when all GT sheep are penned (all modes) ---
# The dog isn't a Supervisor so it can't call simulationQuit() —
# instead we write a sentinel file the launcher polls for and uses
# to kill the Webots process. Bounded by `_gt_sheep` so we don't
# Auto-finish: when all GT sheep are penned, write the sentinel.
# The launcher polls for it and closes Webots so batch evals don't
# hang after the task is done. Bounded by `_gt_sheep` so we don't
# fire during the first few steps while the receiver fills.
if _gt_sheep and not os.path.exists(_DAGGER_DONE_FILE):
gt_active_count = sum(1 for x, y in _gt_sheep.values()
if not is_penned_position(x, y))
if gt_active_count == 0:
if MODE == "dagger":
_dump_dagger_log()
os.makedirs(os.path.dirname(_DAGGER_DONE_FILE), exist_ok=True)
open(_DAGGER_DONE_FILE, "w").close()
if _gt_sheep and not _run_done:
gt_active = sum(1 for x, y in _gt_sheep.values()
if not is_penned_position(x, y))
if gt_active == 0:
os.makedirs(os.path.dirname(RUN_DONE_FILE), exist_ok=True)
open(RUN_DONE_FILE, "w").close()
_run_done = True
print(f"[dog] all {len(_gt_sheep)} sheep penned at step "
f"{step_count} — wrote {_DAGGER_DONE_FILE}, "
f"launcher will close Webots")
if MODE == "dagger" and step_count % DAGGER_FLUSH_STEPS == 0 and DAGGER_LOG_OBS:
_dump_dagger_log()
f"{step_count} — wrote sentinel, launcher will close Webots")
if step_count % 200 == 0:
gt_penned = sum(1 for x, y in _gt_sheep.values()
if is_penned_position(x, y))
gt_total = len(_gt_sheep)
extra = ""
if MODE == "dagger":
extra = f" logged={len(DAGGER_LOG_OBS)}"
print(f"[dog mode={MODE}] step={step_count} "
f"GT_penned={gt_penned}/{gt_total} "
f"tracks_active={tracker.n_active()} "
f"tracks_penned={tracker.n_penned()} "
f"detections={len(detections)} action=({vx:+.2f}, {vy:+.2f}){extra}")
# Loop ended (Webots told us to quit). Flush any remaining DAgger log.
_dump_dagger_log()
f"detections={len(detections)} action=({vx:+.2f}, {vy:+.2f})")