Cartographer源码阅读2D&3D-PGM生成及地图发布

Cartographer源码阅读2D&3D-PGM生成及地图发布

pgm生成

cartographer提供了pbstream转pgm的service,当然,也可以改写相关函数,直接从cartographer的后端取出数据生成pgm,但是需要修改相关代码。将pbstream转为pgm的节点在pbstream_to_ros_map_main.cc文件里实现。主要函数为:

void Run(const std::string& pbstream_filename, const std::string& map_filestem,
         const double resolution) {
  ::cartographer::io::ProtoStreamReader reader(pbstream_filename);
  ::cartographer::io::ProtoStreamDeserializer deserializer(&reader);
  // 从pbstream中读取pose graph
  const auto& pose_graph = deserializer.pose_graph();

  LOG(INFO) << "Loading submap slices from serialized data.";
  // 存放submap的slice
  std::map<::cartographer::mapping::SubmapId, ::cartographer::io::SubmapSlice>
      submap_slices;
  ::cartographer::mapping::proto::SerializedData proto;
  while (deserializer.ReadNextSerializedData(&proto)) {
    // 只取出submap
    if (proto.has_submap()) {
      const auto& submap = proto.submap();
      // 获取id
      const ::cartographer::mapping::SubmapId id{
          submap.submap_id().trajectory_id(),
          submap.submap_id().submap_index()};
      // 获取global pose
      const ::cartographer::transform::Rigid3d global_submap_pose =
          ::cartographer::transform::ToRigid3(
              pose_graph.trajectory(id.trajectory_id)
                  .submap(id.submap_index)
                  .pose());
      // 填充该submap的slice
      FillSubmapSlice(global_submap_pose, submap, &submap_slices[id]);
    }
  }
  CHECK(reader.eof());

  LOG(INFO) << "Generating combined map image from submap slices.";
  // 画上网格
  auto result =
      ::cartographer::io::PaintSubmapSlices(submap_slices, resolution);

  ::cartographer::io::StreamFileWriter pgm_writer(map_filestem + ".pgm");

  ::cartographer::io::Image image(std::move(result.surface));
  // 生成pgm
  WritePgm(image, resolution, &pgm_writer);

  const Eigen::Vector2d origin(
      -result.origin.x() * resolution,
      (result.origin.y() - image.height()) * resolution);

  ::cartographer::io::StreamFileWriter yaml_writer(map_filestem + ".yaml");
  // 生成yaml
  WriteYaml(resolution, origin, pgm_writer.GetFilename(), &yaml_writer);
}

submapQuery.srv和SubmapTexture.msg

SubmapQuery:
int32 trajectory_id
int32 submap_index
---
cartographer_ros_msgs/StatusResponse status
int32 submap_version // submap包含的node的数量
cartographer_ros_msgs/SubmapTexture[] textures
    
SubmapTexture:
// 存放的有cell的value和alpha值
uint8[] cells
int32 width
int32 height
float64 resolution
geometry_msgs/Pose slice_pose

submap slice:

using UniqueCairoSurfacePtr =
    std::unique_ptr<cairo_surface_t, void (*)(cairo_surface_t*)>;
UniqueCairoSurfacePtr MakeUniqueCairoSurfacePtr(cairo_surface_t* surface) {
  return UniqueCairoSurfacePtr(surface, cairo_surface_destroy);
}

struct SubmapSlice {
  SubmapSlice()
      : surface(::cartographer::io::MakeUniqueCairoSurfacePtr(nullptr)) {}

  // Texture data.
  int width;
  int height;
  int version;
  double resolution;
  // global pose
  ::cartographer::transform::Rigid3d slice_pose;
  // 网格数据
  ::cartographer::io::UniqueCairoSurfacePtr surface;
  // Pixel data used by 'surface'. Must outlive 'surface'.
  // 网格数据转换成pgm需要的格式
  std::vector<uint32_t> cairo_data;

  // Metadata.
  // local pose网格在submap坐标系下的位姿
  ::cartographer::transform::Rigid3d pose;
  int metadata_version = -1;
};

struct SubmapTexture {
  struct Pixels {
    std::vector<char> intensity;
    std::vector<char> alpha;
  };
  Pixels pixels;
  int width;
  int height;
  double resolution;
  ::cartographer::transform::Rigid3d slice_pose;
};

slice填充:

// 填充slice
void FillSubmapSlice(
    const ::cartographer::transform::Rigid3d& global_submap_pose,
    const ::cartographer::mapping::proto::Submap& proto,
    SubmapSlice* const submap_slice) {
  // 生成SubmapQuery 的response
  ::cartographer::mapping::proto::SubmapQuery::Response response;
  ::cartographer::transform::Rigid3d local_pose;
  if (proto.has_submap_3d()) {
    mapping::Submap3D submap(proto.submap_3d());
    local_pose = submap.local_pose();
    submap.ToResponseProto(global_submap_pose, &response);
  } else {
    ::cartographer::mapping::Submap2D submap(proto.submap_2d());
    // local pose 后面没用到啊
    local_pose = submap.local_pose();
    submap.ToResponseProto(global_submap_pose, &response);
  }
  // slice的global pose
  submap_slice->pose = global_submap_pose;
  // 取出texture
  auto& texture_proto = response.textures(0);
  // 解压得到cell存放的数值:value和alpha
  const SubmapTexture::Pixels pixels = UnpackTextureData(
      texture_proto.cells(), texture_proto.width(), texture_proto.height());
  // 设置宽
  submap_slice->width = texture_proto.width();
  // 设置高
  submap_slice->height = texture_proto.height();
  // 设置分辨率
  submap_slice->resolution = texture_proto.resolution();
  // 设置local pose
  submap_slice->slice_pose =
      ::cartographer::transform::ToRigid3(texture_proto.slice_pose());
  // 生成surface 和 cairo_data
  submap_slice->surface =
      DrawTexture(pixels.intensity, pixels.alpha, texture_proto.width(),
                  texture_proto.height(), &submap_slice->cairo_data);
}

// 生成SubmapQuery 的response
void Submap2D::ToResponseProto(
    const transform::Rigid3d&,
    proto::SubmapQuery::Response* const response) const {
  if (!grid_) return;
  // 填充submap数量
  response->set_submap_version(num_range_data());
  // 填充texture
  proto::SubmapQuery::Response::SubmapTexture* const texture =
      response->add_textures();
  // 将网格数据放入texture中
  grid()->DrawToSubmapTexture(texture, local_pose());
}

// 网格数据放入texture中
bool ProbabilityGrid::DrawToSubmapTexture(
    proto::SubmapQuery::Response::SubmapTexture* const texture,
    transform::Rigid3d local_pose) const {
  // 裁剪网格
  Eigen::Array2i offset;
  CellLimits cell_limits;
  ComputeCroppedLimits(&offset, &cell_limits);

  std::string cells;
  // 遍历所有的网格,生成网格的string
  for (const Eigen::Array2i& xy_index : XYIndexRangeIterator(cell_limits)) {
    if (!IsKnown(xy_index + offset)) {
      cells.push_back(0 /* unknown log odds value */);
      cells.push_back(0 /* alpha */);
      continue;
    }
    // We would like to add 'delta' but this is not possible using a value and
    // alpha. We use premultiplied alpha, so when 'delta' is positive we can
    // add it by setting 'alpha' to zero. If it is negative, we set 'value' to
    // zero, and use 'alpha' to subtract. This is only correct when the pixel
    // is currently white, so walls will look too gray. This should be hard to
    // detect visually for the user, though.
    const int delta =
        128 - ProbabilityToLogOddsInteger(GetProbability(xy_index + offset));
    const uint8 alpha = delta > 0 ? 0 : -delta;
    const uint8 value = delta > 0 ? delta : 0;
    cells.push_back(value);
    cells.push_back((value || alpha) ? alpha : 1);
  }
  // 压缩string
  common::FastGzipString(cells, texture->mutable_cells());
  // 设置宽
  texture->set_width(cell_limits.num_x_cells);
  // 设置高
  texture->set_height(cell_limits.num_y_cells);
  // 设置分辨率
  const double resolution = limits().resolution();
  texture->set_resolution(resolution);
  // 设置Local pose,该local pose是网格在submap坐标系下的位姿
  const double max_x = limits().max().x() - resolution * offset.y();
  const double max_y = limits().max().y() - resolution * offset.x();
  *texture->mutable_slice_pose() = transform::ToProto(
      local_pose.inverse() *
      transform::Rigid3d::Translation(Eigen::Vector3d(max_x, max_y, 0.)));

  return true;
}

// 生成surface 和 cairo_data
UniqueCairoSurfacePtr DrawTexture(const std::vector<char>& intensity,
                                  const std::vector<char>& alpha,
                                  const int width, const int height,
                                  std::vector<uint32_t>* const cairo_data) {
  CHECK(cairo_data->empty());

  // Properly dealing with a non-common stride would make this code much more
  // complicated. Let's check that it is not needed.
  const int expected_stride = 4 * width;
  CHECK_EQ(expected_stride, cairo_format_stride_for_width(kCairoFormat, width));
  for (size_t i = 0; i < intensity.size(); ++i) {
    // We use the red channel to track intensity information. The green
    // channel we use to track if a cell was ever observed.
    const uint8_t intensity_value = intensity.at(i);
    const uint8_t alpha_value = alpha.at(i);
    const uint8_t observed =
        (intensity_value == 0 && alpha_value == 0) ? 0 : 255;
    // 逐个将cell数据生成uint32_t数据,放入cairo_data中
    cairo_data->push_back((alpha_value << 24) | (intensity_value << 16) |
                          (observed << 8) | 0);
  }
  // 构造UniqueCairoSurfacePtr
  auto surface = MakeUniqueCairoSurfacePtr(cairo_image_surface_create_for_data(
      reinterpret_cast<unsigned char*>(cairo_data->data()), kCairoFormat, width,
      height, expected_stride));
  CHECK_EQ(cairo_surface_status(surface.get()), CAIRO_STATUS_SUCCESS)
      << cairo_status_to_string(cairo_surface_status(surface.get()));
  return surface;
}

画网格:

struct PaintSubmapSlicesResult {
  PaintSubmapSlicesResult(::cartographer::io::UniqueCairoSurfacePtr surface,
                          Eigen::Array2f origin)
      : surface(std::move(surface)), origin(origin) {}
  ::cartographer::io::UniqueCairoSurfacePtr surface;

  // Top left pixel of 'surface' in map frame.
  Eigen::Array2f origin;
};

PaintSubmapSlicesResult PaintSubmapSlices(
    const std::map<::cartographer::mapping::SubmapId, SubmapSlice>& submaps,
    const double resolution) {
  // 计算所有的submap的外包矩形
  Eigen::AlignedBox2f bounding_box;
  {
    auto surface = MakeUniqueCairoSurfacePtr(
        cairo_image_surface_create(kCairoFormat, 1, 1));
    auto cr = MakeUniqueCairoPtr(cairo_create(surface.get()));
    const auto update_bounding_box = [&bounding_box, &cr](double x, double y) {
      cairo_user_to_device(cr.get(), &x, &y);
      bounding_box.extend(Eigen::Vector2f(x, y));
    };

    CairoPaintSubmapSlices(
        1. / resolution, submaps, cr.get(),
        [&update_bounding_box](const SubmapSlice& submap_slice) {
          update_bounding_box(0, 0);
          update_bounding_box(submap_slice.width, 0);
          update_bounding_box(0, submap_slice.height);
          update_bounding_box(submap_slice.width, submap_slice.height);
        });
  }
  // 两边分别扩展5个Pixel
  const int kPaddingPixel = 5;
  // 得到新的box
  const Eigen::Array2i size(
      std::ceil(bounding_box.sizes().x()) + 2 * kPaddingPixel,
      std::ceil(bounding_box.sizes().y()) + 2 * kPaddingPixel);
  // 计算新的box的中心
  const Eigen::Array2f origin(-bounding_box.min().x() + kPaddingPixel,
                              -bounding_box.min().y() + kPaddingPixel);
  // 画上网格数据
  auto surface = MakeUniqueCairoSurfacePtr(
      cairo_image_surface_create(kCairoFormat, size.x(), size.y()));
  {
    auto cr = MakeUniqueCairoPtr(cairo_create(surface.get()));
    cairo_set_source_rgba(cr.get(), 0.5, 0.0, 0.0, 1.);
    cairo_paint(cr.get());
    cairo_translate(cr.get(), origin.x(), origin.y());
    CairoPaintSubmapSlices(1. / resolution, submaps, cr.get(),
                           [&cr](const SubmapSlice& submap_slice) {
                             cairo_set_source_surface(
                                 cr.get(), submap_slice.surface.get(), 0., 0.);
                             cairo_paint(cr.get());
                           });
    cairo_surface_flush(surface.get());
  }
  return PaintSubmapSlicesResult(std::move(surface), origin);
}

void CairoPaintSubmapSlices(
    const double scale,
    const std::map<::cartographer::mapping::SubmapId, SubmapSlice>& submaps,
    cairo_t* cr, std::function<void(const SubmapSlice&)> draw_callback) {
  cairo_scale(cr, scale, scale);

  for (auto& pair : submaps) {
    // 获取submap slice
    const auto& submap_slice = pair.second;
    if (submap_slice.surface == nullptr) {
      return;
    }
    // global pose * local pose
    const Eigen::Matrix4d homo =
        ToEigen(submap_slice.pose * submap_slice.slice_pose).matrix();

    cairo_save(cr);
    cairo_matrix_t matrix;
    cairo_matrix_init(&matrix, homo(1, 0), homo(0, 0), -homo(1, 1), -homo(0, 1),
                      homo(0, 3), -homo(1, 3));
    cairo_transform(cr, &matrix);

    const double submap_resolution = submap_slice.resolution;
    cairo_scale(cr, submap_resolution, submap_resolution);

    // Invokes caller's callback to utilize slice data in global cooridnate
    // frame. e.g. finds bounding box, paints slices.
    draw_callback(submap_slice);
    cairo_restore(cr);
  }
}

生成pgm和yaml:

void WritePgm(const ::cartographer::io::Image& image, const double resolution,
              ::cartographer::io::FileWriter* file_writer) {
  const std::string header = "P5\n# Cartographer map; " +
                             std::to_string(resolution) + " m/pixel\n" +
                             std::to_string(image.width()) + " " +
                             std::to_string(image.height()) + "\n255\n";
  file_writer->Write(header.data(), header.size());
  for (int y = 0; y < image.height(); ++y) {
    for (int x = 0; x < image.width(); ++x) {
      const char color = image.GetPixel(x, y)[0];
      file_writer->Write(&color, 1);
    }
  }
}

void WriteYaml(const double resolution, const Eigen::Vector2d& origin,
               const std::string& pgm_filename,
               ::cartographer::io::FileWriter* file_writer) {
  // Magic constants taken directly from ros map_saver code:
  // https://github.com/ros-planning/navigation/blob/ac41d2480c4cf1602daf39a6e9629142731d92b0/map_server/src/map_saver.cpp#L114
  const std::string output =
      "image: " + pgm_filename + "\n" +
      "resolution: " + std::to_string(resolution) + "\n" + "origin: [" +
      std::to_string(origin.x()) + ", " + std::to_string(origin.y()) +
      ", 0.0]\nnegate: 0\noccupied_thresh: 0.65\nfree_thresh: 0.196\n";
  file_writer->Write(output.data(), output.size());
}

地图发布

地图发布节点在occupancy_grid_node_main.cc文件中,发布的地图的topic名称为/map(在node_constants.h中定义 kOccupancyGridTopic),发布的消息为:nav_msgs::OccupancyGrid,入口函数为HandleSubmapList,其接收到SubmapList消息即发布占据栅格数据。

class Node {
 public:
  // 网格分辨率和发布频率构造Node
  explicit Node(double resolution, double publish_period_sec);
  ~Node() {}

  Node(const Node&) = delete;
  Node& operator=(const Node&) = delete;

 private:
  void HandleSubmapList(const cartographer_ros_msgs::SubmapList::ConstPtr& msg);
  void DrawAndPublish(const ::ros::WallTimerEvent& timer_event);
  void PublishOccupancyGrid(const std::string& frame_id, const ros::Time& time,
                            const Eigen::Array2f& origin,
                            cairo_surface_t* surface);

  ::ros::NodeHandle node_handle_;
  const double resolution_;

  ::cartographer::common::Mutex mutex_;
  ::ros::ServiceClient client_ GUARDED_BY(mutex_);
  ::ros::Subscriber submap_list_subscriber_ GUARDED_BY(mutex_);
  ::ros::Publisher occupancy_grid_publisher_ GUARDED_BY(mutex_);
  // submap id和submap slice组合
  std::map<SubmapId, SubmapSlice> submap_slices_ GUARDED_BY(mutex_);
  ::ros::WallTimer occupancy_grid_publisher_timer_;
  std::string last_frame_id_;
  ros::Time last_timestamp_;
};

void Node::HandleSubmapList(
    const cartographer_ros_msgs::SubmapList::ConstPtr& msg) {
  ::cartographer::common::MutexLocker locker(&mutex_);

  // We do not do any work if nobody listens.
  if (occupancy_grid_publisher_.getNumSubscribers() == 0) {
    return;
  }

  // Keep track of submap IDs that don't appear in the message anymore.
  std::set<SubmapId> submap_ids_to_delete;
  // 上一次发布后的submap slice中所有的submap id放入submap_ids_to_delete
  for (const auto& pair : submap_slices_) {
    submap_ids_to_delete.insert(pair.first);
  }
    
  for (const auto& submap_msg : msg->submap) {
    const SubmapId id{submap_msg.trajectory_id, submap_msg.submap_index};
    // 当前接收到的submap list 中包含该id,则将其从submap_ids_to_delete删除,剩余的为可能要删除的id
    submap_ids_to_delete.erase(id);
    // 获取上次发布的submap slice
    SubmapSlice& submap_slice = submap_slices_[id];
    // 更新global pose
    submap_slice.pose = ToRigid3d(submap_msg.pose);
    // 更新submap 包含的node的数量
    submap_slice.metadata_version = submap_msg.submap_version;
    if (submap_slice.surface != nullptr &&
        submap_slice.version == submap_msg.submap_version) {
      continue;
    }
    // 获取submap slice的texture数据
    auto fetched_textures =
        ::cartographer_ros::FetchSubmapTextures(id, &client_);
    if (fetched_textures == nullptr) {
      continue;
    }
    CHECK(!fetched_textures->textures.empty());
    submap_slice.version = fetched_textures->version;

    // We use the first texture only. By convention this is the highest
    // resolution texture and that is the one we want to use to construct the
    // map for ROS.
    const auto fetched_texture = fetched_textures->textures.begin();
    // slice 数据赋值
    submap_slice.width = fetched_texture->width;
    submap_slice.height = fetched_texture->height;
    submap_slice.slice_pose = fetched_texture->slice_pose;
    submap_slice.resolution = fetched_texture->resolution;
    submap_slice.cairo_data.clear();
    // 更新slice数据的cairo_data和surface
    submap_slice.surface = ::cartographer::io::DrawTexture(
        fetched_texture->pixels.intensity, fetched_texture->pixels.alpha,
        fetched_texture->width, fetched_texture->height,
        &submap_slice.cairo_data);
  }

  // Delete all submaps that didn't appear in the message.
  for (const auto& id : submap_ids_to_delete) {
    submap_slices_.erase(id);
  }

  last_timestamp_ = msg->header.stamp;
  last_frame_id_ = msg->header.frame_id;
}
// 通过服务获取submap slice的texture数据
std::unique_ptr<::cartographer::io::SubmapTextures> FetchSubmapTextures(
    const ::cartographer::mapping::SubmapId& submap_id,
    ros::ServiceClient* client) {
  ::cartographer_ros_msgs::SubmapQuery srv;
  srv.request.trajectory_id = submap_id.trajectory_id;
  srv.request.submap_index = submap_id.submap_index;
  if (!client->call(srv) ||
      srv.response.status.code != ::cartographer_ros_msgs::StatusCode::OK) {
    return nullptr;
  }
  if (srv.response.textures.empty()) {
    return nullptr;
  }
  auto response =
      ::cartographer::common::make_unique<::cartographer::io::SubmapTextures>();
  response->version = srv.response.submap_version;
  for (const auto& texture : srv.response.textures) {
    // 获取cell数据的string
    const std::string compressed_cells(texture.cells.begin(),
                                       texture.cells.end());
    // 解压string,生成cell数据
    response->textures.emplace_back(::cartographer::io::SubmapTexture{
        ::cartographer::io::UnpackTextureData(compressed_cells, texture.width,
                                              texture.height),
        texture.width, texture.height, texture.resolution,
        ToRigid3d(texture.slice_pose)});
  }
  return response;
}

void Node::DrawAndPublish(const ::ros::WallTimerEvent& unused_timer_event) {
  if (submap_slices_.empty() || last_frame_id_.empty()) {
    return;
  }

  ::cartographer::common::MutexLocker locker(&mutex_);
  // 画网格
  auto painted_slices = PaintSubmapSlices(submap_slices_, resolution_);
    
  std::unique_ptr<nav_msgs::OccupancyGrid> msg_ptr = CreateOccupancyGridMsg(
      painted_slices, resolution_, last_frame_id_, last_timestamp_);
  occupancy_grid_publisher_.publish(*msg_ptr);
}

// 生成occupancy grid message
std::unique_ptr<nav_msgs::OccupancyGrid> CreateOccupancyGridMsg(
    const cartographer::io::PaintSubmapSlicesResult& painted_slices,
    const double resolution, const std::string& frame_id,
    const ros::Time& time) {
  auto occupancy_grid =
      ::cartographer::common::make_unique<nav_msgs::OccupancyGrid>();

  const int width = cairo_image_surface_get_width(painted_slices.surface.get());
  const int height =
      cairo_image_surface_get_height(painted_slices.surface.get());
  const ros::Time now = ros::Time::now();

  occupancy_grid->header.stamp = time;
  occupancy_grid->header.frame_id = frame_id;
  occupancy_grid->info.map_load_time = time;
  occupancy_grid->info.resolution = resolution;
  occupancy_grid->info.width = width;
  occupancy_grid->info.height = height;
  occupancy_grid->info.origin.position.x =
      -painted_slices.origin.x() * resolution;
  occupancy_grid->info.origin.position.y =
      (-height + painted_slices.origin.y()) * resolution;
  occupancy_grid->info.origin.position.z = 0.;
  occupancy_grid->info.origin.orientation.w = 1.;
  occupancy_grid->info.origin.orientation.x = 0.;
  occupancy_grid->info.origin.orientation.y = 0.;
  occupancy_grid->info.origin.orientation.z = 0.;

  const uint32_t* pixel_data = reinterpret_cast<uint32_t*>(
      cairo_image_surface_get_data(painted_slices.surface.get()));
  occupancy_grid->data.reserve(width * height);
  for (int y = height - 1; y >= 0; --y) {
    for (int x = 0; x < width; ++x) {
      const uint32_t packed = pixel_data[y * width + x];
      const unsigned char color = packed >> 16;
      const unsigned char observed = packed >> 8;
      const int value =
          observed == 0
              ? -1
              : ::cartographer::common::RoundToInt((1. - color / 255.) * 100.);
      CHECK_LE(-1, value);
      CHECK_GE(100, value);
      occupancy_grid->data.push_back(value);
    }
  }

  return occupancy_grid;
}
  • 3
    点赞
  • 13
    收藏
    觉得还不错? 一键收藏
  • 2
    评论
### 回答1: 在cartographer中,使用2D点云进行扫描匹配时,可以使用ceresscanmatch功能。这个功能是基于Ceres Solver库实现的。Ceres Solver是一个非线性优化库,用于解决各种最小化问题。在cartographer中,ceresscanmatch被用于解决2D点云匹配的问题。 具体来说,ceresscanmatch用于匹配两个相邻帧的2D点云。在进行扫描匹配时,需要先对数据进行滤波处理,然后使用ceres进行优化,找到两个点云之间的最佳匹配。在这个过程中,需要使用一种优化算法来最小化匹配误差,这个误差是通过计算点云之间的距离来得到的。 相比于其他扫描匹配方法,ceresscanmatch的优势在于它能够进行非常精准的匹配。这是因为它使用了一个非线性优化算法,能够处理复杂的误差函数和约束条件。此外,ceresscanmatch还支持使用多种不同的误差函数,以适应不同的应用场景。 总之,ceresscanmatch是cartographer中用于2D点云扫描匹配的一个非常重要的功能,它能够让我们更加准确、稳定地进行扫描匹配,并且支持广泛的应用场景。 ### 回答2: 本文将继续介绍cartographer中的ceres扫描匹配部分,ceres扫描匹配是利用Ceres Solver进行的位姿优化,可以准确估计机器人运动的姿态。 ceres扫描匹配部分主要包括ceres_scan_matcher.cc和ceres_scan_matcher.h两个文件。其中ceres_scan_matcher.cc包含了ceres扫描匹配算法的具体实现,而ceres_scan_matcher.h则是相关的头文件。 ceres_scan_matcher.cc中的函数主要有两个,分别是CeresScanMatcher::Match()和CeresScanMatcher::MatchFullSubmap()。其中,CeresScanMatcher::Match()函数用于实现一次扫描匹配,输入参数为当前激光数据和候选的位姿,输出参数为匹配的位姿和评估值。 在CeresScanMatcher::Match()函数中,先通过叶芽上下文来获取轨迹和submap,然后将当前激光数据转换为点云,并对点云进行滤波和预处理,接着定义优化问题和相关的参数,其中优化问题使用ceres::Problem类来定义,相关参数则定义在CeresScanMatcherOptions结构体中,最后通过ceres::Solve()函数进行位姿优化。 CeresScanMatcher::MatchFullSubmap()函数则用于在整个submap上进行匹配,并返回匹配的位姿和评估值。它的实现与CeresScanMatcher::Match()函数类似,只是输入参数为整个submap的信息。 综上所述,ceres扫描匹配部分利用Ceres Solver进行位姿优化,可以准确估计机器人运动的姿态,是cartographer中重要的功能之一。 ### 回答3: cartographer是一款开源的SLAM系统,其源代码完整透明,方便研究和理解。其中,2D点云扫描匹配是cartographer中的一个重要功能,而这一功能又是由ceres扫描匹配实现的。 ceresscanmatch是cartographer中的一个重要模块,用于实现2D点云的扫描匹配。在这个模块中,ceres solver被用来进行优化过程。具体来说,ceresscanmatch会将已知位姿下的实测点云与预测的点云进行匹配,得到匹配误差。随后,ceres solver会对这些匹配误差进行非线性优化,最终得到最优位姿。这样,就能够实现快速准确的2D点云扫描匹配,从而提高了SLAM系统的性能和精度。 在详细研究ceresscanmatch之前,首先需要了解一下ceres solver。ceres solver是一个基于C++的非线性优化库,用于解决复杂的数值优化问题。在cartographer中,ceres solver被用来进行扫描匹配的优化过程,应用目标函数和求解器来寻求最优解。其中,目标函数是由误差项和状态变量构成的,求解器则用来解决这个目标函数并确定状态变量的最优化值。 具体而言,在cartographer中,扫描匹配的目标函数是根据传感器数据得出的,其包括一系列误差项和参考帧的相对位姿。在每个迭代步骤中,ceres solver会计算目标函数的梯度和海森矩阵,并利用这些值来更新参考帧的位姿。当误差项最小化时,相对位姿就能够得到最优解。 总之,ceresscanmatch是cartographer中的一个重要模块,用于实现2D点云的扫描匹配。借助ceres solver进行优化,可以实现高效准确的扫描匹配,为SLAM系统的实现提供了重要的基础。

“相关推荐”对你有帮助么?

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论 2
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值