5.2.4 Padding

The padding subsystem provides safe, realistic striking targets and embeds impact sensing for training analytics. It must absorb repeated high-velocity strikes without transmitting dangerous forces to internal mechanisms or the user, while allowing embedded IMU sensors to detect each impact.

Requirements & Considerations

This subsystem addresses DO-1 (Performance Analytics) and DO-4 (Adaptive Fight Intelligence). Requirements were derived from the need to absorb repeated punch impacts safely and provide strike-zone discrimination for training analytics.

System Requirements

The following system-level requirement is inherited from the Robot Mechanism (Section 5.2) requirements table and governs all padding design decisions:

Subsystem Acceptance Criteria

The following acceptance criteria were derived from design decisions specific to this subsystem that introduce failure modes not captured by the system-level requirement above. PAD-AC-1 arises because RM-6 states "provide impact detection" but does not specify a minimum detection rate: the choice to use an IMU threshold-based algorithm (L2 norm ≥ 2.0 g) introduces a miss-detection failure mode driven by threshold calibration, sensor placement, and padding damping. A 95% true-positive rate is needed to make training analytics meaningful. PAD-AC-2 arises because RM-6 does not specify that the sensing system must differentiate force intensity: the IMU L2 vector norm response must be monotonically ordered (light < medium < heavy) for the analytics layer to provide meaningful skill-level feedback. This is a data-quality failure mode not captured by any RM code.

Table: Padding Subsystem Requirements and Acceptance Criteria

System Design Narrative

Following the Systems Engineering V-Model introduced in Section 3.2, RM-6, PAD-AC-1, and PAD-AC-2 were fixed before detailed foam selection, sensor placement, or detection algorithm decisions were made. The diagram below applies the V-Model to the Padding subsystem specifically.

Left Side of the V: Design Decomposition

RM-6 (absorb strikes + impact detection) drove the multi-layer padding architecture: polyethylene foam outer layer, anti-vibration isolation mounts to protect the drivetrain, and MPU6050 IMU sensors embedded between the foam surface and the isolation interface. PAD-AC-1 (95% detection rate) drove the IMU threshold calibration (L2 norm ≥ 2.0 g), the dual-bus I²C topology (four sensors across two hardware buses on Teensy 4.0), and the Hard STOP bus-hang recovery protocol. PAD-AC-2 (force differentiation) drove using the L2 norm magnitude as the sensing metric rather than a binary threshold, ensuring graded response across punch intensities.

Base of the V: Integration Build

At integration stage, the foam pads with embedded IMU sensors were mounted to the arm assembly via anti-vibration isolation mounts. Four MPU6050 sensors were wired across two I²C buses on the Teensy 4.0, firmware initialised and confirmed communicating, and the /strike_events ROS 2 topic was verified publishing correctly during initial impact tests.

Right Side of the V: Verification Closure

PAD-AC-1 closes via the 60-punch controlled detection test (≥ 57 punches detected). PAD-AC-2 closes via the L2-norm monotonicity test (3 users × 3 intensities, one-way ANOVA). RM-6 (structural survivability) closes via the 200-punch mechanical endurance test. Full test procedures are in Testing & Evaluation.

Design

Multi-Layer Padding Architecture

A multi-layer construction with anti-vibration isolation mounts separates impact absorption (outer foam) from structural mounting (isolation interface to arm mechanism).

• Outer layer: Polyethylene foam — lightweight, sufficient energy absorption for repeated strikes
• Isolation mounts: Anti-vibration mounts decouple the padding from the 2-DOF arm mechanism, preventing impact vibration from propagating into 3D-printed drivetrain components
• Sensing layer: MPU6050 IMU sensors at Head, Celiac Plexus, and Liver zones — embedded between the foam surface and the isolation interface

The isolation mount is critical because the Arm Actuation subsystem's 3D-printed gears and Delrin alignment pins are optimised for torsional and compressive loads, not repeated transverse impacts. Without mechanical decoupling, user punches to the padding would transmit impulse forces directly into the gear mesh, accelerating fatigue failure.

Low-Level Control — IMU Sensing Pipeline

Quad-IMU I²C Topology

Up to four MPU6050 6-axis IMUs (3-axis accelerometer + 3-axis gyroscope) are distributed across the padding zones to detect and classify physical strikes. The sensor placement enables spatial discrimination — determining not only whether a punch landed, but where on the robot's body it connected.

The MPU6050 supports only two I²C addresses (0x68 and 0x69, selected by the AD0 pin). To accommodate four sensors, the Teensy 4.0's two independent hardware I²C buses are both utilised:

Both buses operate at 400 kHz (I²C Fast Mode). Each MPU6050 read transaction requires approximately 14 bytes, consuming ~280 µs per sensor at 400 kHz.

Strike Detection Pipeline

The raw 3-axis acceleration from each IMU is included in the expanded /motor_feedback payload (21 doubles) published via micro-ROS at 200 Hz. On the application side (the application), the IMU Diagnostics Tab computes the L2 norm of each sensor's acceleration vector:

||a|| = √(a_x² + a_y² + a_z²)

When ||a|| exceeds a configurable threshold (default: 2.0 g), the system registers a valid strike event and publishes the event data (sensor ID, peak acceleration, timestamp) to the /strike_events ROS 2 topic.

Validation

The following table summarises the verification outcome for the padding subsystem, tracing closure back to the Robot Mechanism verification matrix.

Detailed Documentation

For in-depth technical details, refer to the padding-specific documentation:

Cross-References

• 5.2.5 Arm Actuation — The padding is mechanically attached to the 2-DOF arm via anti-vibration isolation mounts. The integration between padding sensing and arm control is discussed at the Robot Mechanism system level. Coordinated strike validation using IMU feedback as input is documented as a validation component under Arm Actuation.