pick_ik_parameters.yaml

pick_ik:
  # Overall solver settings
  mode: {
    type: string,
    default_value: "global",
    description: "IK solver mode. Set to global to allow the initial guess to be a long distance from the goal, or local if the initial guess is near the goal.",
    validation: {
      one_of<>: [["global", "local"]]
    }
  }
  gd_step_size: {
    type: double,
    default_value: 0.0001,
    description: "Gradient descent step size for joint perturbation",
    validation: {
      gt_eq<>: [1.0e-12],
    }
  }
  gd_max_iters: {
    type: int,
    default_value: 100,
    description: "Maximum iterations for local gradient descent",
    validation: {
      gt_eq<>: [1],
    }
  }
  gd_min_cost_delta: {
    type: double,
    default_value: 1.0e-12,
    description: "Minimum change in cost function value for gradient descent",
    validation: {
      gt_eq<>: [1.0e-64],
    }
  }
  # Cost functions and thresholds
  position_threshold: {
    type: double,
    default_value: 0.001,
    description: "Position threshold for solving IK, in meters",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  orientation_threshold: {
    type: double,
    default_value: 0.001,
    description: "Orientation threshold for solving IK, in radians",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  approximate_solution_position_threshold: {
    type: double,
    default_value: 0.05,
    description: "Position threshold for approximate IK solutions, in meters. If displacement is larger than this, the approximate solution will fall back to the initial guess",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  approximate_solution_orientation_threshold: {
    type: double,
    default_value: 0.05,
    description: "Orientation threshold for approximate IK solutions, in radians. If displacement is larger than this, the approximate solution will fall back to the initial guess",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  approximate_solution_joint_threshold: {
    type: double,
    default_value: 0.0,
    description: "Joint threshold for approximate IK solutions, in radians. If displacement is larger than this, the approximate solution will fall back to the initial guess",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  approximate_solution_cost_threshold: {
    type: double,
    default_value: 0.0,
    description: "Cost threshold for approximate IK solutions. If the result of the cost function is larger than this, the approximate solution will fall back to the initial guess.",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  cost_threshold: {
    type: double,
    default_value: 0.001,
    description: "Scalar value for comparing to result of cost functions. IK is considered solved when all position/rotation/twist thresholds are satisfied and all cost functions return a value lower than this value.",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  position_scale: {
    type: double,
    default_value: 1.0,
    description: "The position scale for the pose cost function. Set to 0.0 to solve for only rotation or other goals",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  rotation_scale: {
    type: double,
    default_value: 0.5,
    description: "The rotation scale for the pose cost function. Set to 0.0 to solve for only position",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  center_joints_weight: {
    type: double,
    default_value: 0.0,
    description: "Weight for centering cost function, >0.0 enables const function",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  avoid_joint_limits_weight: {
    type: double,
    default_value: 0.0,
    description: "Weight for avoiding joint limits cost function, >0.0 enables const function",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  minimal_displacement_weight: {
    type: double,
    default_value: 0.0,
    description: "Weight for minimal displacement cost function, >0.0 enables const function",
    validation: {
      gt_eq<>: [0.0],
    },
  }
  stop_optimization_on_valid_solution : {
    type: bool,
    default_value: true,
    description: "If false, keeps running after finding a solution to further optimize the solution until a time or iteration limit is reached",
  }
  # Memetic IK specific parameters
  memetic_num_threads: {
    type: int,
    default_value: 1,
    description: "Number of threads for memetic IK",
    validation: {
      gt_eq<>: [1],
    }
  }
  memetic_stop_on_first_solution: {
    type: bool,
    default_value: true,
    description: "If true, stops on first solution and terminates other threads",
  }
  memetic_population_size: {
    type: int,
    default_value: 16,
    description: "Population size for memetic IK",
    validation: {
      gt_eq<>: [1],
    }
  }
  memetic_elite_size: {
    type: int,
    default_value: 4,
    description: "Number of elite members of memetic IK population",
    validation: {
      gt_eq<>: [1],
    }
  }
  memetic_wipeout_fitness_tol: {
    type: double,
    default_value: 0.00001,
    description: "Minimum fitness must improve by this value or population will be wiped out",
    validation: {
      gt_eq<>: [0.0],
    }
  }
  memetic_max_generations: {
    type: int,
    default_value: 100,
    description: "Maximum iterations of evolutionary algorithm",
    validation: {
      gt_eq<>: [1],
    }
  }
  memetic_gd_max_iters: {
    type: int,
    default_value: 25,
    description: "Maximum iterations of gradient descent during memetic exploitation",
    validation: {
      gt_eq<>: [1],
    }
  }
  memetic_gd_max_time: {
    type: double,
    default_value: 0.005,
    description: "Maximum time spent on gradient descent during memetic exploitation",
    validation: {
      gt_eq<>: [0.0],
    }
  }