distance_approach_ipopt_cuda_interface.h

Go to the documentation of this file.

/******************************************************************************
 * Copyright 2019 The Apollo Authors. All Rights Reserved.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 *****************************************************************************/

/*
 * @file
 */

#pragma once

#include <vector>

#include <omp.h>

#include <adolc/adolc.h>
#include <adolc/adolc_openmp.h>
#include <adolc/adolc_sparse.h>
#include <adolc/adouble.h>

#include <coin/IpTNLP.hpp>
#include <coin/IpTypes.hpp>

#include "Eigen/Dense"

#include "cyber/common/log.h"
#include "cyber/common/macros.h"
#include "modules/common/configs/proto/vehicle_config.pb.h"
#include "modules/common/configs/vehicle_config_helper.h"
#include "modules/common/math/math_utils.h"
#include "modules/common/util/util.h"
#include "modules/planning/common/planning_gflags.h"
#include "modules/planning/open_space/trajectory_smoother/distance_approach_interface.h"
#include "modules/planning/proto/planner_open_space_config.pb.h"

#define tag_f 1
#define tag_g 2
#define tag_L 3
#define HPOFF 30

namespace apollo {
namespace planning {

class DistanceApproachIPOPTCUDAInterface : public DistanceApproachInterface {
 public:
  DistanceApproachIPOPTCUDAInterface(
      const size_t horizon, const double ts, const Eigen::MatrixXd& ego,
      const Eigen::MatrixXd& xWS, const Eigen::MatrixXd& uWS,
      const Eigen::MatrixXd& l_warm_up, const Eigen::MatrixXd& n_warm_up,
      const Eigen::MatrixXd& x0, const Eigen::MatrixXd& xf,
      const Eigen::MatrixXd& last_time_u, const std::vector<double>& XYbounds,
      const Eigen::MatrixXi& obstacles_edges_num, const size_t obstacles_num,
      const Eigen::MatrixXd& obstacles_A, const Eigen::MatrixXd& obstacles_b,
      const PlannerOpenSpaceConfig& planner_open_space_config);

  virtual ~DistanceApproachIPOPTCUDAInterface() = default;

  bool get_nlp_info(int& n, int& m, int& nnz_jac_g, int& nnz_h_lag,  // NOLINT
                    IndexStyleEnum& index_style) override;           // NOLINT

  bool get_bounds_info(int n, double* x_l, double* x_u, int m, double* g_l,
                       double* g_u) override;

  bool get_starting_point(int n, bool init_x, double* x, bool init_z,
                          double* z_L, double* z_U, int m, bool init_lambda,
                          double* lambda) override;

  bool eval_f(int n, const double* x, bool new_x, double& obj_value) override;

  bool eval_grad_f(int n, const double* x, bool new_x, double* grad_f) override;
  // eval_grad_f by hand.
  bool eval_grad_f_hand(int n, const double* x, bool new_x, double* grad_f);

  bool eval_g(int n, const double* x, bool new_x, int m, double* g) override;

  bool check_g(int n, const double* x, int m, const double* g);

  bool eval_jac_g(int n, const double* x, bool new_x, int m, int nele_jac,
                  int* iRow, int* jCol, double* values) override;
  // sequential implementation to jac_g
  bool eval_jac_g_ser(int n, const double* x, bool new_x, int m, int nele_jac,
                      int* iRow, int* jCol, double* values);
  // parallel implementation to jac_g
  bool eval_jac_g_par(int n, const double* x, bool new_x, int m, int nele_jac,
                      int* iRow, int* jCol, double* values);

  bool eval_h(int n, const double* x, bool new_x, double obj_factor, int m,
              const double* lambda, bool new_lambda, int nele_hess, int* iRow,
              int* jCol, double* values) override;

  void finalize_solution(Ipopt::SolverReturn status, int n, const double* x,
                         const double* z_L, const double* z_U, int m,
                         const double* g, const double* lambda,
                         double obj_value, const Ipopt::IpoptData* ip_data,
                         Ipopt::IpoptCalculatedQuantities* ip_cq) override;

  void get_optimization_results(Eigen::MatrixXd* state_result,
                                Eigen::MatrixXd* control_result,
                                Eigen::MatrixXd* time_result,
                                Eigen::MatrixXd* dual_l_result,
                                Eigen::MatrixXd* dual_n_result) const;

  //***************    start ADOL-C part ***********************************
  template <class T>
  bool eval_obj(int n, const T* x, T* obj_value);

  template <class T>
  bool eval_constraints(int n, const T* x, int m, T* g);

  void generate_tapes(int n, int m, int* nnz_h_lag);
  //***************    end   ADOL-C part ***********************************

 private:
  int num_of_variables_ = 0;
  int num_of_constraints_ = 0;
  int horizon_ = 0;
  int lambda_horizon_ = 0;
  int miu_horizon_ = 0;
  double ts_ = 0.0;
  Eigen::MatrixXd ego_;
  Eigen::MatrixXd xWS_;
  Eigen::MatrixXd uWS_;
  Eigen::MatrixXd l_warm_up_;
  Eigen::MatrixXd n_warm_up_;
  Eigen::MatrixXd x0_;
  Eigen::MatrixXd xf_;
  Eigen::MatrixXd last_time_u_;
  std::vector<double> XYbounds_;

  // debug flag
  bool enable_constraint_check_;

  // penalty
  double weight_state_x_ = 0.0;
  double weight_state_y_ = 0.0;
  double weight_state_phi_ = 0.0;
  double weight_state_v_ = 0.0;
  double weight_input_steer_ = 0.0;
  double weight_input_a_ = 0.0;
  double weight_rate_steer_ = 0.0;
  double weight_rate_a_ = 0.0;
  double weight_stitching_steer_ = 0.0;
  double weight_stitching_a_ = 0.0;
  double weight_first_order_time_ = 0.0;
  double weight_second_order_time_ = 0.0;

  double w_ev_ = 0.0;
  double l_ev_ = 0.0;
  std::vector<double> g_;
  double offset_ = 0.0;
  Eigen::MatrixXi obstacles_edges_num_;
  int obstacles_num_ = 0;
  int obstacles_edges_sum_ = 0;
  double wheelbase_ = 0.0;

  Eigen::MatrixXd state_result_;
  Eigen::MatrixXd dual_l_result_;
  Eigen::MatrixXd dual_n_result_;
  Eigen::MatrixXd control_result_;
  Eigen::MatrixXd time_result_;

  // obstacles_A
  Eigen::MatrixXd obstacles_A_;

  // obstacles_b
  Eigen::MatrixXd obstacles_b_;

  // whether to use fix time
  bool use_fix_time_ = false;

  // state start index
  int state_start_index_ = 0;

  // control start index.
  int control_start_index_ = 0;

  // time start index
  int time_start_index_ = 0;

  // lagrangian l start index
  int l_start_index_ = 0;

  // lagrangian n start index
  int n_start_index_ = 0;

  double min_safety_distance_ = 0.0;

  double max_safety_distance_ = 0.0;

  double max_steer_angle_ = 0.0;

  double max_speed_forward_ = 0.0;

  double max_speed_reverse_ = 0.0;

  double max_acceleration_forward_ = 0.0;

  double max_acceleration_reverse_ = 0.0;

  double min_time_sample_scaling_ = 0.0;

  double max_time_sample_scaling_ = 0.0;

  double max_steer_rate_ = 0.0;

  double max_lambda_ = 0.0;

  double max_miu_ = 0.0;

 private:
  DistanceApproachConfig distance_approach_config_;
  PlannerOpenSpaceConfig planner_open_space_config_;
  const common::VehicleParam vehicle_param_ =
      common::VehicleConfigHelper::GetConfig().vehicle_param();

 private:
  //***************    start ADOL-C part ***********************************
  double* obj_lam;
  unsigned int* rind_L; /* row indices    */
  unsigned int* cind_L; /* column indices */
  double* hessval;      /* values */
  int nnz_L = 0;
  int options_L[4];
  //***************    end   ADOL-C part ***********************************
};

}  // namespace planning
}  // namespace apollo