L3 Specialist Curriculum

Five sections: multi-platform diagnostics, FFT analysis, fleet diagnostics, MTBF, and FMEA.

Section 1: Multi-Platform Diagnostics

L3 specialists service robots across the full BCR-supported platform catalog. Each platform has unique CAN topology, ROS 2 topic structures, and mechanical characteristics.

Platform Comparison: CAN Topology

Boston Dynamics Spot uses a proprietary internal bus. Field diagnostics use the Spot Python SDK:

from bosdyn.client import create_standard_sdk
sdk = create_standard_sdk('TechMedixDiagnostic')
robot = sdk.create_robot('192.168.80.3')
robot.authenticate('admin', 'password')
state = robot.ensure_client('robot-state').get_robot_state()

Approach Protocol Differences

• Unitree G1/H1-2: 2-meter exclusion zone when powered. Uses onboard voice commands for mode switching.
• Boston Dynamics Spot: 2-meter exclusion zone. Has sit/stand behavior — always command SIT before physical approach.
• DJI Agras T50: 10-meter exclusion zone when armed. Agriculture drone with large propeller span (>2.2m). Must be in LANDED state for battery/maintenance access.

Section 2: FFT Analysis and Bearing Defect Detection

Fast Fourier Transform (FFT) converts a time-domain vibration signal into the frequency domain, revealing periodic fault signatures buried in noise.

Sampling and Nyquist Theorem

• To capture a frequency of f Hz, sample at minimum 2f Hz (Nyquist theorem)
• For bearing defect detection up to 10 kHz, sample accelerometers at minimum 25 kHz
• Higher sampling rates reduce aliasing risk but increase data volume

FFT Bin Resolution

• Frequency resolution = Sample rate / Number of FFT points
• Example: 25,000 Hz sample rate, 4096-point FFT: resolution = 6.1 Hz per bin
• Finer resolution requires more data points (longer capture time)

Bearing Defect Frequency Formula (BPFO)

Ball Pass Frequency Outer Race:

BPFO = (N/2) * (RPM/60) * (1 - (d/D) * cos(alpha))

Where:

• N = number of rolling elements
• d = rolling element diameter (mm)
• D = pitch diameter (mm)
• alpha = contact angle (degrees)

Example (G1 knee bearing: N=14, d=6mm, D=40mm, alpha=15 deg, RPM=300):

• BPFO = (14/2) * (300/60) * (1 - (6/40) * cos(15°))
• BPFO = 7 * 5 * (1 - 0.145)
• BPFO = 35 * 0.855 = 29.9 Hz

A peak at ~30 Hz in the vibration FFT indicates outer race wear on this bearing.

Section 3: Fleet-Level Diagnostics

Fleet Health Dashboard Interpretation

TechMedix fleet health score aggregates:

• Individual robot health scores (weighted by fleet size)
• Active alert pressure (P1 = -10 pts, P2 = -3 pts each)
• SLA compliance rate (late responses reduce score)
• Maintenance currency (robots past service interval reduce score)

Signal Correlation Across Fleet

When multiple robots show the same fault pattern:

• Check if the pattern correlates to a batch of robots from the same production date
• Check if the fault correlates to a recent firmware update
• Check if the fault is concentrated in a specific region (environmental factor)

Use TechMedix alert grouping: Filter > Group by Fault Code to identify fleet-wide patterns.

Section 4: MTBF Analysis

Mean Time Between Failures (MTBF) is the statistical average time between robot failures.

Calculation

MTBF = Total Operating Hours / Number of Failures

Example: 5 robots, each operated for 200 hours in 6 months, with 8 total failures:

• Total operating hours = 5 * 200 = 1,000 hours
• MTBF = 1,000 / 8 = 125 hours

Predictive Maintenance Scheduling

If MTBF = 125 hours and robots operate 40 hours/week:

• Average failure interval = 3.1 weeks
• Schedule preventive maintenance at 80% of MTBF = 100 hours (every 2.5 weeks)

TechMedix automatically calculates fleet MTBF when telemetry is logged consistently. Access via Fleet > Analytics > MTBF Dashboard.

Section 5: Advanced FMEA

FMEA Structure

Failure Mode and Effects Analysis systematically catalogs failure modes and their risk:

Escalation Thresholds

• RPN > 500 OR Severity >= 9: escalate to L4 and BCR engineering immediately
• RPN 200-500: flag for priority maintenance at next window
• RPN < 200: include in standard maintenance backlog

TechMedix FMEA Integration

TechMedix automatically calculates RPN from telemetry:

• Severity: derived from fault impact on operation (P1 = S8-10, P2 = S5-7, P3 = S1-4)
• Occurrence: derived from failure frequency in fleet history
• Detectability: derived from sensor coverage for that failure mode

Navigate to Fleet > FMEA to review current RPN scores for all monitored failure modes.