fix the node manifest and make the example dataflow self-contained
validate / validate (pull_request) Successful in 58m46s
validate / test (pull_request) Successful in 58m47s
validate / markdownlint (pull_request) Successful in 58m43s

Two problems, both found by the validate job once it could finally run.

categories named `teleop` and `input`, neither of which dora defines. The set is
fixed — sensor, actuator, robot, transform, filter, ml-inference, llm, speech,
communication, recorder, visualization, simulator, debug — and an unknown
variant makes the whole manifest unparseable, so every type annotation in it was
being ignored.

dataflow.yml spanned three repositories through sibling paths, which meant dora
could not resolve the other nodes' manifests and reported their typed outputs as
unannotated. It also could not be validated in CI, which clones one repo. The
house pattern is a self-contained dataflow, and this now follows it: teleop plus
a bench stand-in for the arm and the browser, both from this checkout. The real
three-node wiring moves to README.md, where a consumer would look for it.

The bench answers every service, publishes plausible telemetry, and prints the
velocities it is sent, so the node can be driven from a real gamepad with no
robot in the room. It is not a simulator and must not become one: it never
integrates a velocity or moves a pose.
This commit is contained in:
2026-09-15 16:51:30 -05:00
parent 8a7bfb02e0
commit e9fa4bf36b
6 changed files with 204 additions and 45 deletions
+5
View File
@@ -77,6 +77,11 @@ a source goes away**.
first time somebody picks up the pad. Errors name the variable.
- **`dora_teleop/log.py`** — Timestamped stderr logging. stderr because dora reads a
dynamic node's stdout.
- **`examples/bench.py`** — A stand-in for the arm *and* the browser, so `dataflow.yml`
is self-contained and CI can validate it with no sibling clones. **It is not a
simulator** and must not become one: it never integrates a velocity or moves a pose. Its
job is to close the loop so this node can be watched, not to stand in for
`dora-ulite6` — anything needing the arm to actually respond needs the arm.
- **`dora_teleop/probe.py`** — `dora-teleop-probe`: shows the raw axes, the buttons and the
velocity that would be sent, with no dataflow and no arm. It exists because the indices
in `mapping.py` are SDL's documented layout rather than an observed one, and a pad that
+25 -3
View File
@@ -152,7 +152,7 @@ stops the node at startup rather than the first time somebody picks up the pad.
## Usage
See `dataflow.yml` for the whole stack wired together. The short version:
The real stack is three nodes: this one, the WebSocket bridge and the arm.
```yaml
- id: teleop
@@ -172,12 +172,34 @@ See `dataflow.yml` for the whole stack wired together. The short version:
outputs: [jog, request, ui_state]
```
Point `ws-transceiver`'s `data_in` at `teleop/ui_state` and its `STATIC_DIR` at this
repo's `web/`, and wire `ulite6`'s `request` and `jog` from this node.
**Requirements:** dora 1.0+ (verified with `dora-cli 1.0.1`), Python 3.11+, `uv`, and a
gamepad if you want one — the node starts without and says so.
### Running it with no arm and no browser
`dataflow.yml` in this repo is **not** that stack. It is this node plus a bench stand-in,
so you can drive the real thing from a real gamepad with no robot in the room:
```bash
dora run dataflow.yml --uv
```
**Requirements:** dora 1.0+ (verified with `dora-cli 1.0.1`), Python 3.11+, `uv`, and a
gamepad if you want one — the node starts without and says so.
The bench answers every service, publishes plausible telemetry, and prints each velocity
it is sent:
```text
jog 1 [ +38.89, +23.53, +0.00, +0.00, +0.00, +0.00]
jog 2 [ +38.89, +23.53, +0.00, +0.00, +0.00, +0.00]
jog 3 [ +0.00, +0.00, +0.00, +0.00, +0.00, +0.00] <- stop
```
It is **not a simulator**: it never integrates the velocity or moves the pose, and a jog
is printed and forgotten. It closes the loop so this node's behaviour can be watched —
the lease changing hands, the deadman firing when you let go — not so the arm can be
pretended at.
## Development
+21 -41
View File
@@ -1,12 +1,13 @@
# The whole operator stack, and what CI validates.
# The repo's own example, and what CI validates.
#
# Built from checkouts rather than tags, so `dora validate` resolves each node's
# manifest from its `path:` and checks the wiring against the versions in front
# of you. The sibling repos are expected next to this one, which is how they are
# cloned in practice.
# Self-contained on purpose: teleop plus a bench stand-in for the arm and the
# browser, both built from this checkout, so `dora validate` resolves every
# manifest from `path:` and CI needs no gamepad, no robot and no sibling clones.
#
# Three nodes, one loop: the bridge carries the browser, this node decides who
# drives, and the arm node moves the arm and reports back.
# This is **not** the real stack. The bench answers services and publishes
# telemetry but never moves anything — it exists so the node can be driven from
# a real gamepad with no arm in the room. README.md carries the three-node
# wiring a consumer actually uses.
strict_types: true
@@ -16,13 +17,11 @@ nodes:
build: pip install .
inputs:
tick: dora/timer/millis/50
ui_command: ws-transceiver/data_out
clients: ws-transceiver/clients
response: ulite6/response
motion_state: ulite6/motion_state
robot_state: ulite6/robot_state
tcp_pose: ulite6/tcp_pose
joint_pose: ulite6/joint_pose
clients: bench/clients
response: bench/response
robot_state: bench/robot_state
tcp_pose: bench/tcp_pose
joint_pose: bench/joint_pose
outputs:
- jog
- request
@@ -31,35 +30,16 @@ nodes:
SPEED_MM_S: "60"
SPEED_DEG_S: "30"
- id: ws-transceiver
path: ../dora-websocket-transceiver
build: pip install .
- id: bench
path: examples/bench.py
inputs:
data_in: teleop/ui_state
outputs:
- data_out
- clients
env:
WS_PORT: "9000"
STATIC_DIR: "../dora-teleop/web"
HTTP_PORT: "9010"
- id: ulite6
path: ../dora-ulite6
build: pip install .
inputs:
tick: dora/timer/millis/100
request: teleop/request
tick: dora/timer/millis/200
jog: teleop/jog
request: teleop/request
ui_state: teleop/ui_state
outputs:
- response
- motion_state
- joint_pose
- tcp_pose
- robot_state
env:
ROBOT_IP: "192.168.1.192"
TOOL_TYPE: "vacuum"
# Nothing moves because the node started. An operator asks for the first
# motion, which is the whole point of a teleop stack.
INIT_MODE: "none"
- tcp_pose
- joint_pose
- clients
+1 -1
View File
@@ -15,7 +15,7 @@ apiVersion: 1
name: dora-teleop
namespace: liria
description: Hands a robot arm to a human, from a gamepad or a browser
categories: [robot, teleop, input]
categories: [robot]
keywords: [teleop, gamepad, joystick, jog, arm, manipulator, web-console, operator]
runtime: python
+110
View File
@@ -0,0 +1,110 @@
"""A bench stand-in for the arm and the browser, so teleop can run with neither.
Run by `dataflow.yml`. It answers every service, publishes plausible telemetry,
and prints the velocities it is sent — which is enough to drive the node from a
real gamepad and watch what a real arm would have been told to do.
**It is not a simulator.** It does not integrate the velocity, move a pose, or
model the arm in any way; a jog is printed and forgotten. Its job is to close
the loop so the node's own behaviour can be watched, not to stand in for
`dora-ulite6`. Anything that needs the arm to actually respond needs the arm.
"""
import json
import pyarrow as pa
from dora import Node
REQUEST_ID = "request_id"
# Somewhere in front of the robot, tool pointing down. Fixed, because a moving
# pose here would look like the arm was responding, which it is not.
POSE = [250.0, 0.0, 200.0, 180.0, 0.0, 0.0]
JOINTS = [0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
def main() -> None:
"""Answer whatever teleop asks and report what it would have moved."""
node = Node("bench")
announced = False
tool_engaged = False
freedrive = False
jogs = 0
shown = None
for event in node:
if event["type"] != "INPUT":
continue
if event["id"] == "tick":
if not announced:
announced = True
node.send_output("robot_state", pa.array(["ready"]))
node.send_output("tcp_pose", pa.array(POSE, type=pa.float32()), {"units_position": "mm"})
node.send_output("joint_pose", pa.array(JOINTS, type=pa.float32()), {"units": "deg"})
# No browser is attached to this dataflow, and saying so is what
# keeps teleop from believing one went away mid-jog.
node.send_output("clients", pa.array([{"connected": 0, "event": "close"}]))
elif event["id"] == "jog":
velocity = event["value"][0].as_py()["velocity"]
jogs += 1
moving = any(velocity)
shown = ", ".join(f"{v:+7.2f}" for v in velocity)
print(f"jog {jogs:5d} [{shown}] {'' if moving else '<- stop'}", flush=True)
elif event["id"] == "request":
row = event["value"][0].as_py()
service = row.get("service", "")
request_id = event["metadata"].get(REQUEST_ID, "")
if service == "grasp":
tool_engaged = True
elif service == "release":
tool_engaged = False
elif service == "freedrive":
freedrive = bool(row.get("enabled", True))
if service != "get_state":
print(f"service {service} ({request_id})", flush=True)
node.send_output(
"response",
pa.array([_answer(service, tool_engaged, freedrive)]),
{REQUEST_ID: request_id},
)
elif event["id"] == "ui_state":
# No browser is attached, so this is where the console's picture of
# the arm would go. Printing it on change rather than on every tick
# makes the things worth watching — the lease moving, the arm
# starting and stopping — visible without 20 lines a second.
snapshot = json.loads(event["value"][0].as_py())
if snapshot.get("type") != "telemetry":
continue
current = (snapshot["arm_state"], snapshot["owner"], snapshot["estop"])
if current != shown:
shown = current
state, owner, estop = current
print(
f"console would show: {state}, owner {owner}" + (", E-STOPPED" if estop else ""),
flush=True,
)
def _answer(service: str, tool_engaged: bool, freedrive: bool) -> dict:
"""Build the response row the arm node would have sent."""
if service == "get_state":
return {
"ok": True,
"service": service,
"message": "ready",
"connected": True,
"ready": True,
"tool_engaged": tool_engaged,
"freedrive": freedrive,
"pose": POSE,
"joints": JOINTS,
}
return {"ok": True, "service": service, "message": f"{service} accepted by the bench"}
if __name__ == "__main__":
main()
+42
View File
@@ -0,0 +1,42 @@
# Manifest for the bench stand-in.
#
# dora resolves a node's manifest from its `path:`, so without this file the
# bench would be validated against the teleop node's own contract and every port
# would look wrong.
apiVersion: 1
name: bench
namespace: liria
description: Bench stand-in for the arm and the browser, so teleop can run with neither
runtime: python
entrypoint: bench.py
dora: ">=1.0"
inputs:
tick:
description: Drives the telemetry the arm node would publish.
type: std/core/v1/UInt64
jog:
description: A velocity teleop wants the arm to move at. Printed, then forgotten.
request:
description: A service call. Every one is answered ok.
ui_state:
description: >-
The console snapshot teleop publishes. Accepted and dropped: no browser is
attached to this dataflow.
outputs:
response:
description: One answer per request, under the id it arrived with.
robot_state:
description: Reports `ready` once, so teleop stops drawing the arm as disconnected.
type: std/core/v1/String
tcp_pose:
description: A fixed tool-tip pose. Fixed on purpose — nothing here integrates a velocity.
type: std/core/v1/Float32
joint_pose:
description: Fixed joint angles, for the same reason.
type: std/core/v1/Float32
clients:
description: Always zero connected, since the WebSocket bridge is not in this dataflow.