PTankTrack is a physics-driven tank track system for Unreal Engine 5, built on Chaos Vehicle with a custom Verlet + Position-Based Dynamics (PBD) chain solver. It delivers a fully deformable track belt that follows the terrain, with independent suspension, road-wheel collision, and differential drive — going beyond traditional modeled (skinned) or spline-based tracks, which only fake motion through scrolling textures or static spline layouts.
KEY FEATURES
• Custom Chain Physics — the track belt is a 2D particle chain (Verlet integration + PBD distance constraints), fully independent of Chaos.
• Independent Suspension — each road wheel uses a torsion-bar approximation (spring + damping + Sphere Sweep ground detection).
• Road-Wheel Collision — PBD distance constraints keep adjacent road wheels from interpenetrating, with ground-priority and angle-limit awareness.
• Deformable Track — the belt passively follows wheels and terrain through one-way collision, with no feedback loop for natural stability.
• Instanced Track Plates — plates render as a single draw call with correct per-instance motion vectors (no TAA/TSR ghosting at speed), scrolling along the closed-loop spline every tick.
• Differential Drive — Override-mode torque control with forward torque, dual differential steering (wheel differential + body yaw), reverse-as-brake, and handbrake.
REQUIREMENTS
• Unreal Engine 5.5+ (recompile the source for other UE5 versions) • ChaosVehiclesPlugin + EnhancedInput (enabled automatically) • C++ enabled (compiled binaries + source included)
CONTENT
• C++ plugin (runtime) • Example tank model and demo map • English & Chinese README
WHO IS THIS FOR
• Tank, military, and tracked-vehicle games • Developers learning or extending physics-driven track systems • Any project that needs deformable tracks with real suspension
NOTE
PTankTrack is fully functional and ready to use. Because the track is simulated by a CPU-driven particle chain, scenes with a large number of tracked vehicles benefit from performance optimizations such as LOD.