Foam Construction

The striking surface uses polyethylene foam wrapped in water-resistant faux leather. The design draws from commercial boxing focus mitts, which use multi-layer EVA and PU foam stacks to balance responsiveness with cushioning. For BoxBunny, a simpler single-material foam was sufficient because the spring return mechanism (described below) handles the rebound rather than the foam itself. Each pad weighs approximately 150 g. The faux leather outer layer protects the foam from sweat and surface wear during extended training sessions.

Strike-Zone Placement

Iteration 5 fixed the final mount geometry using the revised ergonomic chain in human_ergonomics_analysis_revised.md. The design uses pronasale (nose tip), subxiphoid (epigastric region), and right and left costal-margin flank targets in boxing guard stance (Drillis and Contini, 1966; Gordon et al., 2014).

Target Population and Guard Model

The design target covers users from 150 to 190 cm. Final pad centres are set from the mean of four profile calculations (150F, 150M, 190F, 190M) after applying boxing-guard posture correction. This keeps the fixed frame practical for the whole range while preserving realistic body-zone order in stance.

Revised Ergonomic Outputs Used for CAD

The final robot uses the revised mean guard values: head pad centre at 127.3 cm, epigastric pad centre at 96.4 cm, and both side pads at 93.4 cm. This gives a head-to-epigastric gap of 30.9 cm and a fixed epigastric-to-side-pad offset of 3.0 cm downward (Drillis and Contini, 1966; Gordon et al., 2014).

Pad Size and Coverage with 3.0 cm Off-Centre

The final body pad face remains 180 mm by 130 mm. With a 3.0 cm centre offset, the side pads are clearly lower than the middle pad but still overlap in usable strike area during motion. This avoids a dead vertical strip while preserving the intended hook target level.

Head Pad Sizing and Separation

The head pad remains 200 mm by 230 mm. The revised chain gives a 30.9 cm centre-to-centre gap from head pad to middle body pad in boxing guard, which matches the practical verification range (Pheasant and Haslegrave, 2006).

Why Side Pads Are Off-Centred by 3.0 cm

The side targets are anchored to the costal-margin flank zone used in the revised ergonomics analysis. In this model, the flank strike band is 3.0 cm below the subxiphoid reference. Therefore, both side pads are mounted 3.0 cm below the middle pad centreline by design, not by tuning.

The left side uses the same height as the right side for symmetric hook training and repeatable mounting geometry. In oral terms, the rule is simple: set the middle pad first, then drop each side pad exactly 3.0 cm.

With 180 mm body pads (9.0 cm half-height), the 3.0 cm off-centre creates a controlled vertical overlap between middle and side targets. That gives a continuous strike corridor for combinations while still making the side pads read as a lower hook band.

Pad Edge Contact and Spring Mechanism Clearance

The 130 mm pad width was chosen so that in the natural extended position the upper edge of each liver or stomach-side pad just meets the lower edge of the epigastric pad. There is no gap between adjacent pads at rest. When a pad is struck, the spring mechanism compresses it backward. As it moves back, the contacting edges separate and the pad retracts freely without the face of one pad bearing against the face of an adjacent pad. All three body pads are independently spring-loaded, so each retracts and returns independently. The edge contact at rest is a design choice to avoid visible gaps in the strike surface while still allowing each pad to travel its full spring stroke without interference.

Final Pad Centre Heights

Mount Design Evolution

The padding mount went through five iterations, all 3D-printed in PLA and attached to the robot's 80x80 mm aluminium V-slot extrusion frame.

Spring Calculations

Spring selection evolved in two stages. Iteration 3 started with a 1 kg design assumption and sag-control priority. Iterations 4 and 5 retained the same spring but shifted priority to impact absorption after the enclosed box guide removed sag as a practical risk.

Stage 1: Initial Sag-Driven Requirement (Iteration 3)

Before the box housing was introduced, the mount was treated as a partially unsupported horizontal spring system. The conservative assumption was an end-effector mass of 1.0 kg. Static load is:

F_static = m × g = 1.0 × 9.81 = 9.81 N

For candidate spring rates, static axial sag follows Hooke's law:

x_sag = F / k

Using a 1 kg load case, the design-space checks are:

k = 33 N/mm → x_sag = 9.81/33 = 0.30 mm

k = 20 N/mm → x_sag = 9.81/20 = 0.49 mm

k = 10 N/mm → x_sag = 9.81/10 = 0.98 mm

This is why the early spring shortlist sat in the 25 to 30 N/mm range: lower sag but noticeably stiffer punch feel.

Stage 2: Measured Mass and Impact-Response Tuning (Iterations 4 to 5)

Prototype weighing showed the moving padding assembly was about 150 to 200 g, not 1 kg. Using a midpoint of 0.175 kg:

F_static = 0.175 × 9.81 = 1.72 N

x_sag = 1.72 / 8 = 0.21 mm

This sag is negligible. Also, in Iterations 4 and 5, the spring sits in a boxed guide that constrains lateral droop. So practical spring selection moved from sag prevention to impact feel and frame-load reduction.

Why 8 N/mm Is Safe for Iterations 4 and 5

For the selected spring rate, compression under representative punch loads is:

F = 200 N → x = 200 / 8 = 25.0 mm

F = 300 N → x = 300 / 8 = 37.5 mm

F = 400 N → x = 400 / 8 = 50.0 mm

F = 500 N → x = 500 / 8 = 62.5 mm

These values are in a usable stroke range for a guided return mechanism. The spring is compliant enough to absorb impact instead of passing a sharp impulse to the frame. Relative to a nominal 140 mm free length, the 500 N check gives 62.5 mm compression, which is about 45% of free length. This supports the 8 N/mm choice as compliant but still controllable in the current enclosure.

In short, 8 N/mm is safe because static sag is negligible for the measured mass, the boxed guide controls droop, and dynamic stroke remains meaningful without immediate bottom-out in the intended use range.

Future Softer Spring Direction

Since sag is no longer a governing requirement in Iterations 4 and 5, future revisions can test softer springs mainly for feel tuning. For reference, at 500 N: k = 5 N/mm gives 100 mm compression, while k = 4 N/mm gives 125 mm compression. These can improve forgiveness but need stroke and coil-bind checks in the same box geometry.

The shorter free length (140 mm vs 180 mm) also reduces the moment arm during robot rotation, keeping rotational inertia lower for faster turning response during sparring.

Future Work: TPU Filament

A potential future improvement is to replace the separate foam, spring, and PLA mount with a single-piece TPU (thermoplastic polyurethane) print. TPU is a flexible elastomer that could act as the striking surface, the return mechanism, and the structural mount simultaneously.

Horizontal Honeycomb Concept

A promising approach is a horizontally oriented honeycomb lattice printed in TPU. When a honeycomb structure is rotated 90 degrees so the hexagonal tubes run horizontally, it becomes very rigid along the tube axis (resisting lateral sway) but compresses like an accordion when pushed from the striking direction. This produces unidirectional dampening: high absorption in the punch direction with zero lateral compliance. This property is well suited to a padding mount that must absorb frontal impacts without swaying sideways.

Manufacturing Considerations

TPU print speeds must be kept to 15 to 30 mm/s to avoid extrusion inconsistencies, making large parts time-intensive. Manufacturing defects such as stringing, bulging, and buckling become more pronounced with larger unit cell sizes in open-cell topologies, while closed-cell designs such as honeycombs produce significantly fewer defects and superior load resistance (Di Caprio et al., 2025).

Moisture Sensitivity

TPU is hygroscopic, meaning it absorbs moisture from the environment during printing. Extended print durations increase the exposure window, and absorbed moisture causes polymer chains to break down during heating, degrading both surface finish and mechanical strength (Di Caprio et al., 2025). For a part of this size, a dry-box filament feed system would be necessary.

Future Work: Variable Head-Torso Ratio

The current design uses a fixed 30 cm head-torso separation derived from mean anthropometric proportions. In practice, the ratio between head and torso height is not constant across statures: shorter individuals tend to have a proportionally smaller head-torso gap than taller individuals (approximately 26 cm at 150 cm versus 33 cm at 190 cm). A future revision could introduce a second degree of freedom, for example a secondary linear actuator or adjustable mounting rail, to allow the head-torso spacing to scale with user height. This would improve anatomical accuracy at the extremes of the operating range, at the cost of additional mechanical complexity and calibration.

References

• Di Caprio, F., Franchitti, S., Borrelli, R., Ciminello, M., Morano, C. and Greco, A. (2025). Design of Flexible TPU-Based Lattice Structures for 3D Printing: A Comparative Analysis of Open-Cell Versus Closed-Cell Topologies. Polymers, 17(10), 1299.
• Pheasant, S. and Haslegrave, C.M. (2006). Bodyspace: Anthropometry, Ergonomics and the Design of Work, 3rd ed. CRC Press.
• Drillis, R. and Contini, R. (1966). Body Segment Parameters. Technical Report No. 1166-03, School of Engineering and Science, New York University.
• Gordon, C.C. et al. (2014). 2012 Anthropometric Survey of U.S. Army Personnel: Methods and Summary Statistics. Report No. NATICK/TR-15/007.
• Pierce, J.D., Reinbold, K.A., Lyngard, B.C., Goldman, R.J. and Pastore, C.M. (2006). Direct Measurement of Punch Force During Six Professional Boxing Matches. Journal of Quantitative Analysis in Sports, 2(2).