Illustrated C/C++ source sheets, symbol index, API relationship graph, compiler console and debug trace

LIVE AI / FLEXIBLE SENSING / FLIGHT CONTEXT

Sensoraia API

Live AI and sensing, connected to the flight.
See, listen, inspect and return supported results while airborne.

Loading...
Searching...
No Matches
dipla_gp_belief_planner.cpp File Reference

Finite-grid GP observations, uncertainty updates, and reference route scoring. More...

#include "dipla_gp_belief_planner.hpp"
#include <algorithm>
#include <cmath>
#include <limits>
#include <map>
#include <stdexcept>
#include <utility>

Namespaces

namespace  aipp

Functions

double aipp::fov_radius (double altitude, double angle=60.0)
 Compute the ground-plane footprint radius from altitude and full FOV angle.
int aipp::resolution_block_size (double altitude)
 Select the research sensor's altitude-dependent aggregation block size.
double aipp::noise_model (double altitude, double minimum_altitude=10.0)
 Compute modeled observation variance at an altitude.
Matrix aipp::initialize_covariance (const Grid &grid, double sigma2, double lengthscale)
 Initialize a dense Matérn-3/2 GP covariance matrix.
Sensor aipp::build_sensor (const Grid &grid, const Point &pose, double angle=60.0, double observation_variance=-1.0)
 Build one block-averaged sensor footprint at a pose.
Sensor aipp::future_sensor (const Grid &grid, const std::vector< Point > &path, double angle=60.0, int stride=lookahead_stride, int max_measurements=lookahead_measurement_budget, double observation_variance=-1.0)
 Approximate future measurements along a path under a bounded budget.
Vector aipp::observe (const Sensor &sensor, const Vector &field)
 Apply the sensor's block averages to a full-grid field.
Vector aipp::sample_sensor_noise (const Sensor &sensor, std::mt19937_64 &generator)
 Draw independent zero-mean noise using each sensor row's variance.
std::vector< bool > aipp::importance_mask (const Vector &mean, const Matrix &covariance, const PlannerOptions &options)
 Select cells whose confidence bounds satisfy the utility threshold.
double aipp::expected_reduction (const Matrix &covariance, const std::vector< bool > &mask, const Sensor &sensor)
 Sum expected marginal-variance reduction over important cells.
void aipp::condition (Matrix &covariance, Vector *mean, const Sensor &sensor, const Vector *measurement)
 Apply a Gaussian linear-observation update in place.
Point aipp::clip_point (const Grid &grid, const Point &point, const PlannerOptions &options)
 Clamp a candidate to the map safety inset and altitude interval.
std::vector< Point > aipp::lattice (const Grid &grid, const PlannerOptions &options)
 Generate the deduplicated coarse-to-fine candidate lattice.
std::vector< Point > aipp::spline (const Point &start, const std::vector< Point > &waypoints, int samples_per_segment=spline_samples_per_segment)
 Interpolate controls with a natural cubic spline parameterized by distance.
double aipp::trajectory_objective (const Grid &grid, const Matrix &covariance, const std::vector< bool > &mask, const Point &start, const std::vector< Point > &raw_waypoints, const PlannerOptions &options)
 Score a trajectory by safety violation and distance-normalized information gain.
std::vector< Point > aipp::grid_search (const Grid &grid, const Vector &mean, const Matrix &covariance, const Point &start, const PlannerOptions &options)
 Greedily select a bounded route by information gain per travel cost.

Detailed Description

Finite-grid GP observations, uncertainty updates, and reference route scoring.

Despite the historical filename, this file does not train the ONNX denoiser. It maintains and queries a Gaussian-process (GP) belief about a scalar field, builds synthetic sensor operators, and provides the greedy/spline machinery used by the explicitly selected comparison planner.

The shared linear-observation model
For N grid cells and M sensor rows, each row of H uniformly averages its listed cells. We store those indices rather than a mostly zero M-by-N matrix. Sigma is the dense N-by-N covariance; R is diagonal independent sensor noise.
P = H * Sigma (M by N)
S = H * Sigma * H^T + R (M by M)
B = solve(S, P) (M by N)
posterior_mean = mean + B^T * (measurement - H * mean)
posterior_covariance = Sigma - P^T * B
Expected information gain needs the covariance update but not future measured values. That distinction lets look-ahead reduce hypothetical uncertainty without handing the planner the hidden field's answer sheet.
When changing numerical behavior
The Eigen branch below and the default aigenlib kernels implement the same valid-input model. Keep operator weights, grid order, solver cutoffs, and reduction rules aligned, and run differential tests when changing either. Public contracts stay in the header; implementation notes here explain why.
Warning
Grid construction caps N, not whole-process memory. One 2601-by-2601 double covariance is about 51.6 MiB; hypothetical copies, M-by-N workspaces, and solver allocations are additional. Some helpers assume valid inputs rather than checking every precondition in both numerical backends.
See also
dipla_gp_belief_planner.hpp Public geometry, sensor, and planner API.