Skip to content

README.md

Latest commit

 

History

History
103 lines (75 loc) · 3.32 KB

README.md

File metadata and controls

103 lines (75 loc) · 3.32 KB

Standalone Robotics Examples

Copy-paste-ready Python scripts for integrating libredgetpu modules into robotics platforms (Raspberry Pi, Jetson, etc.). Each script demonstrates a real-time webcam loop with full parameter control via argparse.

Requirements

pip install libredgetpu opencv-python
# Edge TPU USB accelerator required for all examples except --help

Quick Start

cd examples/

# Visual servoing — track brightest spot
python spot_tracker_example.py --image-size 64 --variant bright

# Collision avoidance with time-to-contact
python looming_detector_example.py --tau-threshold 1.2 --ttc-window 10

# Global optical flow
python optical_flow_example.py --pooled --pool-factor 2

# Yaw/heading estimation
python visual_compass_example.py --fov-deg 90 --pooled

# Template matching
python pattern_tracker_example.py --kernel-size 16 --template logo.png

# Matrix multiply benchmark (no webcam)
python matmul_engine_example.py --dim 256 --iterations 1000 --verify

# Echo state network with webcam input
python reservoir_computer_example.py --dim 256 --spectral-radius 0.95

# Place recognition / similarity search
python embedding_similarity_example.py --dim 256 --top-k 3

# ML inference (classification, detection, segmentation, pose)
python simple_invoker_example.py --model detection --score-threshold 0.3

Overview

Script Module Use Case Webcam Latency
spot_tracker_example.py SpotTracker Visual servoing Yes ~1 ms
looming_detector_example.py LoomingDetector Collision avoidance + TTC Yes ~1.5 ms
optical_flow_example.py OpticalFlow Ego-motion estimation Yes ~2 ms
visual_compass_example.py VisualCompass Yaw/heading estimation Yes ~2 ms
pattern_tracker_example.py PatternTracker Template matching Yes ~5 ms
matmul_engine_example.py MatMulEngine Matrix multiply benchmark No ~0.28 ms
reservoir_computer_example.py ReservoirComputer Echo state network Yes ~0.6 ms
embedding_similarity_example.py EmbeddingSimilarity Place recognition Yes ~0.28 ms
simple_invoker_example.py SimpleInvoker ML inference (5 models) Yes 4-15 ms

Common Arguments

All webcam-based scripts share these arguments (via _common.py):

Argument Default Description
--camera 0 Camera device index
--width 640 Camera frame width
--height 480 Camera frame height
--no-display off Headless mode (no OpenCV window)

Headless Mode

For embedded systems without a display:

python spot_tracker_example.py --no-display
# Metrics printed to terminal via single-line \r updates

Keyboard Controls

Key Action
q Quit
r Reset yaw (visual_compass)
t Re-capture template from center (pattern_tracker)
s Snapshot to gallery (embedding_similarity)
c Clear gallery (embedding_similarity)

Adapting for Your Robot

Each script follows the same pattern:

with Module.from_template(...) as module:
    for frame in webcam_loop:
        result = module.compute(frame)
        # result → your robot's control loop

Replace the webcam loop with your camera driver and the display calls with your robot's actuator commands.