Abstract: This paper presents a specialized motion prediction model for the complex, coordinated movements characteristic of dragon dance performance, aiming to enhance fluidity and aesthetic appeal through accurate motion forecasting. The model employs a multi-objective optimization approach, utilizing the Salp Swarm Algorithm (SSA) to effectively capture and predict intricate movement sequences. Through detailed analysis of temporal and spatial parameters inherent in dragon dance, the model can address the challenges posed by the dance’s dynamic and collective motion requirements. Experimental results verify the model’s predictive accuracy and adaptability, showing a significant improvement in forecast precision over traditional methods. These findings highlight the model’s capability to handle complex motion interactions, thereby contributing to research in both motion prediction and performance modeling. Furthermore, this approach offers broader applicability, as the insights gained here may benefit various domains where accurate, real-time movement prediction is essential, including robotics, sports analysis, and animation.
Introduction
The "Bench Dragon" dance is a traditional folk activity where participants carry dozens or even hundreds of decorated wooden benches connected head-to-tail, forming a massive, articulated "dragon." Coordinating this movement is a significant challenge in kinematics: the dragon can be hundreds of meters long, and participants must maintain fluidity and safety while performing tight turns. Existing path-planning methods often struggle with the "head-turning" phase, where the dragon reverses direction, often resulting in discontinuous trajectories or segment collisions. This research seeks to create a model that ensures smooth, continuous motion that is both visually impressive and mathematically safe.
Methodology
The researchers decomposed the movement into three distinct phases: ingress (traveling in), turning around, and egress (traveling out).
Helical Geometry and Kinematics
The primary path was modeled as an Archimedean solenoid (spiral). This provides polar coordinates that ensure uniform scaling and smooth "wrap-around" paths. To solve the head-turning problem, a centrosymmetric S-shaped curve was introduced, consisting of two tangent circular arcs where the radius of the first arc is exactly twice that of the second. This geometry allows for a seamless transition between the coiling-in and coiling-out spirals.
Multi-Objective Optimization
The model balances two competing goals: minimizing the performance footprint to improve the visual spectacle and minimizing collision risk by ensuring non-adjacent segments maintain a safe distance. To find the optimal speed and path, the team utilized an Improved Sparrow Search Algorithm (SSA). This bio-inspired optimization mimics the foraging and anti-predatory behavior of sparrows, offering superior global search capabilities and faster convergence compared to standard algorithms like Grey Wolf Optimization (GWO).
Simulation and Parameters
The study relies on simulated data based on the physical parameters of a real-world bench dragon:
- Dragon Head: 341 cm long.
- Body/Tail Segments: 220 cm long each.
- Bench Width: 30 cm.
- Spiral Pitch: 1.7 meters.
- Turnaround Space Radius: 4.5 meters.
Key Insights and Results
A critical finding was the formalization of the relationship between the radius of curvature and segment velocity. As the dragon moves into tighter inner circles, segment velocity must decrease to maintain constant angular velocity, preventing a "whip-crack" effect. The centrosymmetric structure for the turnaround ensures the path is unique and fixed, removing guesswork for performers.
Key results include:
- Optimized Speed: The optimal maximum traveling speed for the dragon head was determined to be 1.878 m/s.
- Stability: The algorithm successfully kept the speed of all body segments below 2 m/s, ensuring team stability.
- Verification: Simulation tables confirmed precise coordinates and velocity magnitudes, proving the dragon could maintain tight formation without collisions.
Limitations and Future Work
While robust, this model is currently a 2D kinematic simulation. It does not account for 3D dynamics (lifting/tilting) or the varying physical stamina of human participants. Future research will explore applications in multi-agent robotic swarms, autonomous vehicle coordination in tight urban spaces, and procedural animation for long, articulated creatures in digital media.
Summary updated