Problem Modelling

The general computing flow is as follows.

Tsb is the transform from sensor to base frame. Or in other words: The sensor pose relative to the robot base. The map can be either a pointer to an EmbreeMap or OptixMap. The Prefix of the Simulator is either Embree for CPU computation or Optix for GPU computation. The suffix of the Simulator is dependent on which sensor model you want to simulate. A few examples:

• SphereSimulatorEmbree - Simulate a velodyne on CPU
• PinholeSimulatorOptix- Simulate a depth camera on GPU
• O1DnSimulatorOptix - Simulate a custom O1DnModel on GPU

Example 1: Simulate 1000 3D LiDaRs on CPU

Now we want to construct the following pipeline.

#include <rmagine/simulation/SphereSimulatorEmbree.hpp>

using namespace rmagine;

SphereSimulatorEmbreePtr construct_simulator(std::string path_to_mesh)
{
    // Default construct the SphereSimulatorEmbree as shared pointer
    SphereSimulatorEmbreePtr sim = std::make_shared<SphereSimulatorEmbree>();

    EmbreeMapPtr map = import_embree_map(path_to_mesh);
    sim->setMap(map);

    // Define sensor model
    SphericalModel model;
    // TODO: fill with model specific parameters
    sim->setModel(model);

    // Set static transform between sensor and base (optional)
    Transform Tsb;
    Tsb.setIdentity();
    sim->setTsb(Tsb);

    return sim;
}

int main(int argc, char** argv)
{
    // Load map and set map pointer to simulator
    std::string path_to_mesh = argv[1];
    SphereSimulatorEmbreePtr sim = construct_simulator(path_to_mesh);

    // Define 1000 poses to simulate from
    Memory<Transform, RAM> poses(1000);
    for(int i = 0; i<poses.size(); i++)
    {
        poses[i].setIdentity();
    }

    // add your desired attributes at intersection here
    using ResultT = Bundle<
        Ranges<RAM> 
    >;

    // Result: Simulate Ranges
    ResultT res = sim->simulate<ResultT>(poses_);
    // res.ranges holds a buffer to the ranges

    return 0;
}

Example 2: Simulate 1000 LiDaRs on GPU

Now we want to construct the following pipeline.

The green cells are memory objects on GPU as you see in the following code snippet.

#include <rmagine/simulation/SphereSimulatorOptix.hpp>

using namespace rmagine;

SphereSimulatorOptixPtr construct_simulator(std::string path_to_mesh)
{
    // Default construct the SphereSimulatorEmbree as shared pointer
    SphereSimulatorOptixPtr sim = std::make_shared<SphereSimulatorOptix>();

    OptixMapPtr map = import_optix_map(path_to_mesh);
    sim->setMap(map);

    // Define sensor model
    SphericalModel model;
    // TODO: fill with model specific parameters
    sim->setModel(model);

    // Set static transform between sensor and base (optional)
    Transform Tsb;
    Tsb.setIdentity();
    sim->setTsb(Tsb);

    return sim;
}

int main(int argc, char** argv)
{
    // Load map and set map pointer to simulator
    std::string path_to_mesh = argv[1];
    SphereSimulatorOptixPtr sim = construct_simulator(path_to_mesh);

    // Define 1000 poses to simulate from
    Memory<Transform, RAM> poses(1000);
    for(int i = 0; i<poses.size(); i++)
    {
        poses[i].setIdentity();
    }

    // upload from CPU to GPU
    Memory<Transform, VRAM_CUDA> poses_ = poses;

    // add your desired attributes at intersection here
    using ResultT = Bundle<
        Ranges<VRAM_CUDA> 
    >;

    // Result: Simulate Ranges
    ResultT res = sim->simulate<ResultT>(poses_);

    // download from GPU to CPU
    // or use CUDA buffer for other computations
    Memory<float, RAM> ranges = res.ranges;

    return 0;
}

Example 3: Simulate 1000 LiDaRs on GPU and Images on CPU

Now we want to construct the following pipeline.

#include <rmagine/simulation/SphereSimulatorOptix.hpp>
#include <rmagine/simulation/PinholeSimulatorEmbree.hpp>

using namespace rmagine;

int main(int argc, char** argv)
{
    // Load map and set map pointer to simulator
    std::string path_to_mesh = argv[1];

    // CONSTRUCTION PART

    // Define Simulators
    SphereSimulatorOptix lidar_sim_gpu;
    PinholeSimulatorEmbree dcam_sim_cpu;

    // Load and set maps
    OptixMapPtr map_gpu = import_optix_map(path_to_mesh);
    EmbreeMapPtr map_cpu = import_embree_map(path_to_mesh);
    lidar_sim_gpu.setMap(map_gpu);
    dcam_sim_cpu.setMap(map_cpu);


    SphericalModel lidar_model;
    PinholeModel dcam_model;
    // TODO: Define models
    lidar_sim_gpu.setModel(lidar_model);
    dcam_sim_cpu.setModel(dcam_model);

    // Define static transforms (optional)
    Transform T_lidar_base;
    Transform T_dcam_base;
    lidar_sim_gpu.setTsb(T_lidar_base);
    dcam_sim_cpu.setTsb(T_dcam_base);

    // SIMULATION PART

    Memory<Transform, RAM> poses(1000);
    // TODO: fill poses

    // upload from CPU to GPU
    Memory<Transform, VRAM_CUDA> poses_ = poses;


    // Simulate Depth cameras ranges on CPU
    using ResultT_RAM = Bundle<
        Ranges<RAM> 
    >;
    ResultT_RAM dcam_res
        = dcam_sim_cpu.simulate<ResultT_RAM>(poses);

    // Simulate LiDaRs ranges on GPU
    using ResultT_VRAM = Bundle<
        Ranges<VRAM_CUDA> 
    >;
    ResultT_VRAM lidar_res
        = lidar_sim_gpu.simulate<ResultT_VRAM>(poses_);

    // Download lidar ranges
    Memory<float, RAM> lidar_ranges = lidar_res.ranges;

    // Results are in dcam_res.ranges and lidar_ranges

    return 0;
}