Hide and Seek Belief Simulation
This interactive simulation demonstrates belief-based algorithms for robot search and tracking in hide-and-seek scenarios. The seeker robot must find a hider using probabilistic reasoning under uncertainty, implementing methods from my 2013 conference paper and 2017 PhD thesis.
- Compare greedy frontier exploration vs. probabilistic belief-based planning
- Visualize POMCP (Bayesian grid) and Particle Filter belief representations
- Design custom maps and control the seeker manually
- See how algorithms handle moving targets and unknown maps
How It Works
Belief Visualization Modes
Relative (Heatmap): Colors are normalized to the highest current probability. Red means "most likely spot compared to others". This mode is useful for seeing belief diffusion and how probability spreads over time.
Absolute (0-1): Strict probability scale where white is 0.0 and red is 1.0. Initially, the map will appear mostly white because probability is spread thin across many cells (e.g., 1/500 = 0.002).
1. Greedy (Frontier Exploration)
The robot maintains a map of Known/Unknown cells. In Smart Mode, it acts as a frontier explorer: it paths to the nearest unknown cell to maximize information gain. This is a deterministic, non-probabilistic approach.
2. POMCP Belief (Bayesian Grid)
Implements a Bayesian grid-based belief update. The robot maintains a probability distribution over all free cells. In Smart Mode, it paths to the cell with the highest probability value (Pmax).
3. Particle Filter
Sample-based belief representation using particles. Each particle represents a hypothesis about the hider's location. Particles are density-estimated into a probability grid [0.0, 1.0] for visualization. In Smart Mode, the robot paths to the area with the highest density of particles.