KNNModel.hpp

/***********************************************************************
Copyright (c) 2017, Shushman Choudhury
All rights reserved.
Authors: Shushman Choudhury <shushmanchoudhury@gmail.com>

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are
met:
  Redistributions of source code must retain the above copyright
  notice, this list of conditions and the following disclaimer.
  Redistributions in binary form must reproduce the above copyright
  notice, this list of conditions and the following disclaimer in the
  documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*************************************************************************/

#ifndef BATCHING_POMP_CSPACEBELIEF_KNNMODEL_HPP_
#define BATCHING_POMP_CSPACEBELIEF_KNNMODEL_HPP_

#include <functional>
#include <exception>
#include <memory>
#include <cmath>
#include <ompl/datastructures/NearestNeighbors.h>
#include <ompl/datastructures/NearestNeighborsLinear.h>
#include <ompl/datastructures/NearestNeighborsGNAT.h>
#include <Eigen/Dense>
#include "Model.hpp"
#include "BeliefPoint.hpp"

namespace batching_pomp {
namespace cspacebelief {


class KNNModel : public virtual Model<BeliefPoint>
{
public:

  KNNModel(size_t _KNN,
           double _prior, double _priorWeight,
           const std::function<double(const BeliefPoint&, const BeliefPoint&)>& _distanceFunction)
  : mNumPoints{0}
  , mKNN{_KNN}
  , mPrior{_prior}
  , mPriorWeight{_priorWeight}
  , mDistanceFunction{_distanceFunction}
  {
    mBeliefPointNN.reset(new ompl::NearestNeighborsGNAT<BeliefPoint>());
    mBeliefPointNN->setDistanceFunction(mDistanceFunction);
  }

  // Setters
  void setPrior(double _prior)
  {
    mPrior = _prior;
  }

  void setPriorWeight(double _priorWeight)
  {
    mPriorWeight = _priorWeight;
  }

  void setKNN(size_t _knn)
  {
    mKNN = _knn;
  }


  // Overriden methods
  void addPoint(const BeliefPoint& data) override
  {
    mBeliefPointNN->add(data);
    mNumPoints++;
  }

  void removePoint(const BeliefPoint& data) override
  {
    mBeliefPointNN->remove(data);
    mNumPoints--;
  }

  double estimate(const BeliefPoint& query) const override
  {
    // This is important for cases where edges need to be initialized before any checks.
    if(mNumPoints == 0) {
      return mPrior;
    }

    size_t knn = std::min(mKNN , mNumPoints);

    std::vector<BeliefPoint> neighboursVect;
    mBeliefPointNN->nearestK(query,knn,neighboursVect);

    if(knn != neighboursVect.size()) {
      throw std::runtime_error("Model could not return the expected number of neighbours");
    }

    Eigen::VectorXd weights(knn);
    Eigen::VectorXd values(knn);

    for(size_t i = 0; i < knn; i++) {
      double distance = std::max(0.0001,mDistanceFunction(query, neighboursVect[i]));
      weights[i] = 1.0/distance;
      values[i] = neighboursVect[i].getValue(); // Second element of pair is value
    }
    double result{mPrior};

    result = weights.dot(values) / weights.sum();

    // Do (result + pw*p) / (1 + pw) for smoothing by prior
    result = (result + mPriorWeight*mPrior) / (1 + mPriorWeight);
    

    return result;
  }

private:
  
  // The current number of belief points in the model  
  size_t mNumPoints;

  // The value of k for the kNN lookup
  size_t mKNN;

  // The prior probability in (0,1) to use for smoothing
  double mPrior;

  // The weight in (0,1) to assign to the prior for smoothing
  double mPriorWeight;

  // The kind of nearest neighbour structure for the belief model, GNAT in this case.
  std::unique_ptr<ompl::NearestNeighbors<BeliefPoint>> mBeliefPointNN;

  // The distance metric for the nearest neighbour structure
  std::function<double(const BeliefPoint&, const BeliefPoint&)> mDistanceFunction;
};

} //namespace cspacebelief
} //namespace batching_pomp

#endif //BATCHING_POMP_CSPACEBELIEF_KNNMODEL_HPP_