Other Bayesian Localization Tricks:

• Gaussian Mixture Models

• SLAM

  • Gaussian slam
  • Density maps

• Possibly Log probabilites for decentralization

  • Send sum of log() of each observed probability so far
  • Receiver has two sums of logs:
    • received observation
    • complete model (received observations + own observations + motion model)
  • On receipt of new sum, complete model += (new - local received sum)
  • Exponentiate sums to find actual probabilities when needed