🎓 Lesson 13
D5
Motion Blur Compensation Using Robot Velocity Feedforward
Motion blur compensation using robot velocity feedforward is a technique that tells the camera in advance how fast the robot is moving, so it can adjust its exposure or image processing to keep the captured images sharp and clear.
🎯 Learning Objectives
- ✓ Explain how robot velocity feedforward reduces motion blur in vision-guided mining robotics
- ✓ Calculate required exposure time reduction based on end-effector linear velocity and pixel resolution
- ✓ Design a feedforward timing interface between a UR10e robot controller and a Basler ace USB3 camera using ROS2 timestamp synchronization
- ✓ Analyze motion blur severity using line-spread function (LSF) metrics from captured images of moving drill-bit markers
📖 Why This Matters
In underground and open-pit mining, robotic drill rigs and autonomous survey platforms must inspect boreholes, verify stemming placement, or guide explosive loading—all while moving. If the onboard vision system captures blurry images due to robot motion, AI-based hole detection fails, leading to misaligned blasts, overbreak, or safety hazards. Motion blur compensation isn’t just about sharper photos—it’s about deterministic timing assurance for mission-critical perception in ISO 13849-1 Category 3 safety loops.
📘 Core Principles
Motion blur arises when scene features move across the camera sensor plane during exposure. In robot-vision integration, two motion sources dominate: (1) ego-motion of the robot-mounted camera, and (2) target motion (e.g., vibrating drill string). Feedforward compensation leverages the robot’s real-time kinematic model—specifically Cartesian end-effector velocity (v_x, v_y, v_z) and angular rates (ω_x, ω_y, ω_z)—to predict pixel displacement. This prediction enables three mitigation strategies: (a) exposure time gating (shutter sync), (b) rolling-shutter skew correction, and (c) inverse filter design in post-processing. Critically, feedforward avoids latency-limited feedback approaches—making it essential for >10 Hz robotic motions typical in drill rig positioning (per ISO 8554-2:2021 imaging performance standards).
📐 Maximum Allowable Exposure Time
To limit motion blur to ≤1 pixel, exposure time must satisfy the constraint derived from projected end-effector velocity onto the image plane. This uses pinhole camera geometry and robot Jacobian mapping.
💡 Worked Example
Problem: A UR10e-mounted Basler acA2440-75um camera (pixel pitch p = 3.45 µm, focal length f = 16 mm) tracks a drill bit moving at v_ee = [0.12, 0.03, 0.0] m/s in base frame. The camera’s optical axis aligns with drill-bit longitudinal axis; J_v maps v_ee to image-plane velocity as [0.085, 0.021] m/s (x,y). What is t_exp_max to hold blur ≤1 pixel?
1.
Step 1: Compute magnitude of image-plane velocity: √(0.085² + 0.021²) = 0.0876 m/s
2.
Step 2: Convert pixel pitch to meters: p = 3.45 × 10⁻⁶ m
3.
Step 3: Apply formula: t_exp_max = (3.45e-6) / (0.0876 × 16e-3) = 3.45e-6 / 1.4016e-3 = 2.46 ms
Answer:
The maximum allowable exposure time is 2.46 ms, which falls within the safe range of 1–5 ms for high-precision borehole centroiding per SME Blasting Best Practices Guide (2023).
🏗️ Real-World Application
At BHP’s Olympic Dam mine, a custom-built robotic drill inspection arm (RIGA-II) uses velocity feedforward to guide stereo-vision-based stemming verification. Robot joint velocities are streamed via EtherCAT to a NI PXIe-8529 vision controller, which computes instantaneous image-plane speed and dynamically sets exposure on two FLIR Blackfly S BFS-U3-16S2C-C cameras. This reduced blur-induced false negatives in stemming gap detection from 11% to <0.8%, directly improving blast fragmentation consistency (measured via WipFrag analysis) and reducing re-drilling costs by AUD $220k/year.
🔧 Interactive Calculator
🔧 Open Industrial Robot Vision Guidance & Integration Calculator📋 Case Connection
📋 Vision-Guided Palletizing Robot for Mixed-SKU E-Commerce Fulfillment
Unstructured tote input with random part orientation, variable box dimensions, and reflective surfaces causing glare
📋 Vision-Guided Deburring Robot for Cast Aluminum Engine Blocks
Variable burr height (0.1–2.4 mm) and location due to inconsistent casting; complex freeform surfaces limiting fixed-too...