Most wheeled robotic platforms operate on horizontal planes where motor loads are symmetric and predictable. In vertical robotics, however, continuous downward gravitational acceleration creates severe torque asymmetries between upward climbing, downward braking, and lateral traversing cycles.
Traditional trapezoidal 6-step commutation generates significant torque ripple and mechanical vibration. When amplified against a thin glass pane, this vibration compromises negative suction stability and causes rapid track slippage. To achieve smooth surface adherence and precise velocity control, modern vertical platforms rely on Field-Oriented Control (FOC) algorithms executed on embedded ARM Cortex or specialized motor-control microcontrollers.
Field-Oriented Control (FOC) Implementation on Vertical Loads
Field-Oriented Control decomposes the stator currents of a 3-phase Brushless DC (BLDC) motor into two decoupled orthogonal vector components:
The Direct axis current (Id), controlling the magnetic flux linkage.
The Quadrature axis current (Iq), controlling the dynamic torque output.
By maintaining Id at zero and precisely modulating Iq via Clarke and Park transformations, the controller achieves smooth, maximum torque-per-ampere output. On a vertical glass pane, this continuous torque modulation minimizes micro-vibrations, ensuring the vacuum sealing lip remains completely flush against the surface at all times.
Sensorless Position Estimation vs. Magnetic Encoders
Determining rotor angle at zero or low speeds is critical when initiating a climb on steep glass:
Hall-effect sensors offer coarse 60-degree electrical resolution, which is insufficient for ultra-smooth low-speed positioning under heavy load.
Modern high-reliability platforms utilize high-resolution magnetic rotary encoders (such as 14-bit absolute angle sensors) communicating over SPI with the motor MCU.
In cost-optimized units utilizing sensorless architectures, microcontrollers deploy High-Frequency Injection (HFI) or sliding mode observers to extract rotor position from Back-EMF signals without requiring physical sensors in the harsh, humid cleaning chamber.
Asymmetric Drive Dynamics and Gravity Compensation
When traversing horizontally across a large glass pane, the upper and lower drive wheels experience radically different physical friction loads:
The lower wheel absorbs the full dynamic weight of the chassis plus tether tension.
The upper wheel tends to experience slight traction reduction due to chassis tilt.
The motion control loop integrates a real-time Gravity Compensation Vector derived from the 6-axis IMU:
Feed-forward torque offsets are calculated continuously at 1 kHz control loop frequencies.
If the optical surface encoder detects micro-slipping (a discrepancy between commanded rotor revolutions and actual displacement), the FOC loop immediately throttles Iq to recover traction before track decoupling occurs.
Empirical Telemetry and Lab Benchmarks
Evaluating motor drive stability requires continuous bench testing under variable friction coefficients, such as transition zones between dry glass, surfactant-treated glass, and framed edge silicone. Key hardware benchmarks include:
Total Harmonic Distortion (THD) in phase currents during abrupt vertical-to-horizontal trajectory switches.
Acoustic noise profiles and decibel ratings under heavy vacuum suction versus low-power traversal.
Real-world mechanical thermal dissipation across H-bridge MOSFETs during continuous 60-minute runs.
For comprehensive hardware teardowns, motor drive architecture comparisons, and empirical bench data across current commercial vertical robotics, explore the technical reports and motor performance datasets published at Window Robot Lab:https://windowrobotlab.com/
Are you implementing Field-Oriented Control (FOC) or sensorless BLDC drivers in your robotics projects? How do you handle torque ripple compensation under dynamic external loads? Share your controller schematics and algorithm insights in the comments below.
Top comments (0)