🦛 Kinematic-Aware Improved Hippo Optimization (IHO) for Robot Path Planning

Official implementation of the paper: "Kinematic-Aware Improved Hippo Optimization with Laplacian Ironing for Swarm-based Path Planning in Cluttered Environments" (under review)

---

🌐 Language

English | 中文

---

💡 Key Contributions

1. Kinematic-Aware Constraint

A kinematic-aware mechanism is embedded into the swarm optimization process to explicitly handle nonholonomic constraints of mobile robots. This eliminates physically infeasible paths (e.g., in-place turns or sharp-angle segments), ensuring that generated trajectories are directly executable on real robotic platforms.

2. Laplacian Ironing Operator

Inspired by geometric signal processing, we propose a Laplacian Ironing Operator that smooths waypoint distributions during late-stage optimization. This operator induces a distinctive cliff-like convergence behavior, significantly improving path smoothness without sacrificing optimality.

---

🏗️ Benchmark Framework

We establish a comprehensive evaluation framework consisting of five challenging environments with varying scales and topologies:

• Small-scale narrow corridor maps ($40 \times 40$)
• Large-scale cluttered maze environments ($80 \times 80$)

All methods are evaluated under a strict collision penalty: $$\lambda_{static} = 10^6$$ ensuring a zero-tolerance safety criterion.

Fig. 1. Multi-dimensional benchmark framework and comparison matrix.

---

---

🎥 Hardware Validation

The proposed method is validated on a real-world mobile robotic platform across multiple complex environments.

Fig. 2. Large-scale constrained maze (Map 5). The mobile robot navigates through a high-occupancy labyrinth using the optimized path from Kinematic-Aware IHO. Note the smooth trajectory in narrow corridors without any physically infeasible sharp turns.

---

📊 Results and Comparisons

We compare IHO with state-of-the-art algorithms including HO (baseline), SBOA, ARO, INFO, PSO, and GWO.

Key Findings:

• 100% collision-free solutions even with small population size ($N = 30$)
• Superior path smoothness and compactness
• Clear late-stage convergence acceleration induced by Laplacian ironing

Map 4 comparison: IHO (blue) achieves smoother paths and exhibits cliff-like convergence behavior.

---

📂 Project Structure

Kinematic-Aware-IHO/
├── src/                    # Core algorithms and environments
│   ├── main.m              # Entry point
│   ├── IHO_Planner.m       # Proposed IHO algorithm
│   ├── HO_Planner.m        # Original HO algorithm
│   └── ...                 # Other baselines (PSO, GWO, etc.)
├── results/                # Generated paths and convergence curves
├── assets/                 # Figures used in the paper
└── hardware_demos/         # Real robot demonstrations

---

⚙️ Requirements

• OS: Windows 10/11, Ubuntu 20.04+, or macOS
• MATLAB: R2023b or later (recommended)
• Toolboxes: None required (fully reproducible using base MATLAB)

---

🚀 Quick Start

git clone https://github.com/Yule-Cai/Kinematic-Aware-IHO.git

Then:

1. Open MATLAB
2. Navigate to the src/ directory
3. Run:

main.m

---

📄 License

This project is licensed under the CC BY-NC-SA 4.0 License.

© 2026 Yule Cai