Examples

Minimal gRPC Control

LuckyRobots is gRPC-only. The most direct way to control a robot is to use LuckyEngineClient to call the unified observation RPC + control RPCs.

import numpy as np
from luckyrobots import LuckyEngineClient

client = LuckyEngineClient(host="127.0.0.1", port=50051)
client.connect()

# Send controls (actuator targets depend on the robot you spawned)
client.send_control(controls=[0.1, 0.2, -0.1], robot_name="two_pandas")

# Read back a unified observation snapshot (AgentService.GetObservation)
obs = client.get_observation(robot_name="two_pandas")
qpos = np.array(list(obs.joint_state.positions))
print("qpos:", qpos)

Using Robot Configuration

Access robot-specific settings for proper action limits:

import numpy as np
from luckyrobots import LuckyRobots

cfg = LuckyRobots.get_robot_config("two_pandas")
limits = cfg["action_space"]["actuator_limits"]
lower = np.array([a["lower"] for a in limits], dtype=np.float32)
upper = np.array([a["upper"] for a in limits], dtype=np.float32)
action = np.random.uniform(lower, upper)

Accessing Observations

from luckyrobots import LuckyEngineClient

client = LuckyEngineClient()
client.connect()
obs = client.get_observation(robot_name="two_pandas")
print("positions:", list(obs.joint_state.positions))
print("velocities:", list(obs.joint_state.velocities))

Resetting Agents

Reset agents during training or control loops:

from luckyrobots import LuckyEngineClient

client = LuckyEngineClient(host="127.0.0.1", port=50051)
client.connect()

# Reset the default agent
resp = client.reset_agent()
if resp.success:
    print(f"Agent reset successful: {resp.message}")

# Reset a specific agent
resp = client.reset_agent(agent_name="agent_0")
if resp.success:
    print(f"Agent 'agent_0' reset successful: {resp.message}")

Reading a LiDAR Scan

Read range scans from a LiDAR sensor. Enable live scanning first, then poll a scan while the simulation is idle (not inside an active step() loop):

from luckyrobots import LuckyEngineClient

client = LuckyEngineClient(host="127.0.0.1", port=50051)
client.connect()
client.wait_for_server(timeout=30.0)

client.set_lidar_live(True)              # fire the lidar every step
scan = client.get_lidar_scan(sensor=0)   # 0-based sensor index
print(scan.beams, "beams")               # channels * azimuth_bins
print(list(scan.ranges)[:8])             # metres per beam; -1.0 = no return

Command Line Usage

Run the included controller example with different options:

# Basic usage (connects to running server, resets agent every 10 seconds)
python examples/controller.py --skip-launch

# Specify robot and scene
python examples/controller.py --robot unitreego1 --scene velocity --task locomotion

# Custom rate and duration
python examples/controller.py --rate 30 --duration 60 --skip-launch

# Custom host/port
python examples/controller.py --host 192.168.1.100 --port 50051 --skip-launch