feat: 🎸 initial commit: forked by gisbi-kim/FAST_LIO_SLAM and add liv…

…ox_ros_driver

.gitignore

@@ -0,0 +1,9 @@
+build
+FAST-LIO/Log/*.png
+FAST-LIO/Log/*.txt
+FAST-LIO/Log/*.csv
+FAST-LIO/Log/*.pdf
+.vscode/c_cpp_properties.json
+.vscode/settings.json
+PCD/*.pcd
+*.pyc

.gitmodules

@@ -0,0 +1,4 @@
+[submodule "include/ikd-Tree"]
+	path = include/ikd-Tree
+	url = https://github.com/hku-mars/ikd-Tree.git
+	branch = fast_lio

FAST-LIO/CMakeLists.txt

@@ -0,0 +1,89 @@
+cmake_minimum_required(VERSION 2.8.3)
+project(fast_lio)
+
+SET(CMAKE_BUILD_TYPE "Debug")
+
+ADD_COMPILE_OPTIONS(-std=c++14 )
+ADD_COMPILE_OPTIONS(-std=c++14 )
+set( CMAKE_CXX_FLAGS "-std=c++14 -O3" )
+
+add_definitions(-DROOT_DIR=\"${CMAKE_CURRENT_SOURCE_DIR}/\")
+
+set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -fexceptions" )
+set(CMAKE_CXX_STANDARD 14)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX_EXTENSIONS OFF)
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++14 -pthread -std=c++0x -std=c++14 -fexceptions")
+
+message("Current CPU archtecture: ${CMAKE_SYSTEM_PROCESSOR}")
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "(x86)|(X86)|(amd64)|(AMD64)" )
+  include(ProcessorCount)
+  ProcessorCount(N)
+  message("Processer number:  ${N}")
+  if(N GREATER 4)
+    add_definitions(-DMP_EN)
+    add_definitions(-DMP_PROC_NUM=3)
+    message("core for MP: 3")
+  elseif(N GREATER 3)
+    add_definitions(-DMP_EN)
+    add_definitions(-DMP_PROC_NUM=2)
+    message("core for MP: 2")
+  else()
+    add_definitions(-DMP_PROC_NUM=1)
+  endif()
+else()
+  add_definitions(-DMP_PROC_NUM=1)
+endif()
+
+find_package(OpenMP QUIET)
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${OpenMP_CXX_FLAGS}")
+set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS}   ${OpenMP_C_FLAGS}")
+
+find_package(PythonLibs REQUIRED)
+find_path(MATPLOTLIB_CPP_INCLUDE_DIRS "matplotlibcpp.h")
+
+find_package(catkin REQUIRED COMPONENTS
+  geometry_msgs
+  nav_msgs
+  sensor_msgs
+  roscpp
+  rospy
+  std_msgs
+  pcl_ros
+  tf
+  livox_ros_driver
+  message_generation
+  eigen_conversions
+)
+
+find_package(Eigen3 REQUIRED)
+find_package(PCL 1.8 REQUIRED)
+
+message(Eigen: ${EIGEN3_INCLUDE_DIR})
+
+include_directories(
+	${catkin_INCLUDE_DIRS} 
+  ${EIGEN3_INCLUDE_DIR}
+  ${PCL_INCLUDE_DIRS}
+  ${PYTHON_INCLUDE_DIRS}
+  include)
+
+add_message_files(
+  FILES
+  Pose6D.msg
+)
+
+generate_messages(
+ DEPENDENCIES
+ geometry_msgs
+)
+
+catkin_package(
+  CATKIN_DEPENDS geometry_msgs nav_msgs roscpp rospy std_msgs message_runtime
+  DEPENDS EIGEN3 PCL
+  INCLUDE_DIRS
+)
+
+add_executable(fastlio_mapping src/laserMapping.cpp include/ikd-Tree/ikd_Tree.cpp src/preprocess.cpp)
+target_link_libraries(fastlio_mapping ${catkin_LIBRARIES} ${PCL_LIBRARIES} ${PYTHON_LIBRARIES})
+target_include_directories(fastlio_mapping PRIVATE ${PYTHON_INCLUDE_DIRS})

FAST-LIO/Log/fast_lio_time_log_analysis.m

@@ -0,0 +1,135 @@
+clear
+close all
+
+Color_red = [0.6350 0.0780 0.1840];
+Color_blue = [0 0.4470 0.7410];
+Color_orange = [0.8500 0.3250 0.0980];
+Color_green = [0.4660 0.6740 0.1880];
+Color_lightblue = [0.3010 0.7450 0.9330];
+Color_purple = [0.4940 0.1840 0.5560];
+Color_yellow = [0.9290 0.6940 0.1250];
+
+fast_lio_ikdtree = csvread("./fast_lio_time_log.csv",1,0);
+timestamp_ikd = fast_lio_ikdtree(:,1);
+timestamp_ikd = timestamp_ikd - min(timestamp_ikd);
+total_time_ikd = fast_lio_ikdtree(:,2)*1e3;
+scan_num = fast_lio_ikdtree(:,3);
+incremental_time_ikd = fast_lio_ikdtree(:,4)*1e3;
+search_time_ikd = fast_lio_ikdtree(:,5)*1e3;
+delete_size_ikd = fast_lio_ikdtree(:,6);
+delete_time_ikd = fast_lio_ikdtree(:,7) * 1e3;
+tree_size_ikd_st = fast_lio_ikdtree(:,8);
+tree_size_ikd = fast_lio_ikdtree(:,9);
+add_points = fast_lio_ikdtree(:,10);
+
+fast_lio_forest = csvread("fast_lio_time_log.csv",1,0);
+fov_check_time_forest = fast_lio_forest(:,5)*1e3;
+average_time_forest = fast_lio_forest(:,2)*1e3;
+total_time_forest = fast_lio_forest(:,6)*1e3;
+incremental_time_forest = fast_lio_forest(:,3)*1e3;
+search_time_forest = fast_lio_forest(:,4)*1e3;
+timestamp_forest = fast_lio_forest(:,1);
+
+% Use slide window to calculate average
+L = 1;     % Length of slide window
+for i = 1:length(timestamp_ikd)
+    if (i<L)
+        average_time_ikd(i) = mean(total_time_ikd(1:i));
+    else
+        average_time_ikd(i) = mean(total_time_ikd(i-L+1:i));
+    end
+end
+for i = 1:length(timestamp_forest)
+    if (i<L)
+        average_time_forest(i) = mean(total_time_forest(1:i));
+    else
+        average_time_forest(i) = mean(total_time_forest(i-L+1:i));
+    end
+end
+
+
+
+
+f = figure;
+set(gcf,'Position',[80 433 600 640])
+tiled_handler = tiledlayout(3,1);
+tiled_handler.TileSpacing = 'compact';
+tiled_handler.Padding = 'compact';
+nexttile;
+hold on;
+set(gca,'FontSize',12,'FontName','Times New Roman') 
+plot(timestamp_ikd, average_time_ikd,'-','Color',Color_blue,'LineWidth',1.2);
+plot(timestamp_forest, average_time_forest,'--','Color',Color_orange,'LineWidth',1.2);
+lg = legend("ikd-Tree", "ikd-Forest",'location',[0.1314 0.8559 0.2650 0.0789],'fontsize',14,'fontname','Times New Roman')
+title("Time Performance on FAST-LIO",'FontSize',16,'FontName','Times New Roman')
+xlabel("time/s",'FontSize',16,'FontName','Times New Roman')
+yl = ylabel("Run Time/ms",'FontSize',15,'Position',[285.7 5.5000 -1]);
+xlim([32,390]);
+ylim([0,23]);
+ax1 = gca;
+ax1.YAxis.FontSize = 12;
+ax1.XAxis.FontSize = 12;
+grid on
+box on
+% print('./Figures/fastlio_exp_average','-depsc','-r600')
+
+
+index_ikd = find(search_time_ikd > 0);
+search_time_ikd = search_time_ikd(index_ikd);
+index_forest = find(search_time_forest > 0);
+search_time_forest = search_time_forest(index_forest);
+
+t = nexttile;
+hold on;
+boxplot_data_ikd = [incremental_time_ikd,total_time_ikd];
+boxplot_data_forest = [incremental_time_forest,total_time_forest];
+Colors_ikd = [Color_blue;Color_blue;Color_blue];
+Colors_forest = [Color_orange;Color_orange;Color_orange];
+% xticks([3,8,13])
+h_search_ikd = boxplot(search_time_ikd,'Whisker',50,'Positions',1,'Colors',Color_blue,'Widths',0.3);
+h_search_forest = boxplot(search_time_forest,'Whisker',50,'Positions',1.5,'Colors',Color_orange,'Widths',0.3);
+h_ikd = boxplot(boxplot_data_ikd,'Whisker',50,'Positions',[3,5],'Colors',Color_blue,'Widths',0.3);
+h_forest = boxplot(boxplot_data_forest,'Whisker',50,'Positions',[3.5,5.5],'Colors',Color_orange,'Widths',0.3);
+ax2 = gca;
+ax2.YAxis.Scale = 'log';
+xlim([0.5,6.0])
+ylim([0.0008,100])
+xticks([1.25 3.25 5.25])
+xticklabels({'Nearest Search','    Incremental Updates','Total Time'});
+yticks([1e-3,1e-2,1e-1,1e0,1e1,1e2])
+ax2.YAxis.FontSize = 12;
+ax2.XAxis.FontSize = 14.5;
+% ax.XAxis.FontWeight = 'bold';
+ylabel('Run Time/ms','FontSize',14,'FontName','Times New Roman')
+box_vars = [findall(h_search_ikd,'Tag','Box');findall(h_ikd,'Tag','Box');findall(h_search_forest,'Tag','Box');findall(h_forest,'Tag','Box')];
+for j=1:length(box_vars)
+    if (j<=3)
+        Color = Color_blue;
+    else
+        Color = Color_orange;
+    end
+    patch(get(box_vars(j),'XData'),get(box_vars(j),'YData'),Color,'FaceAlpha',0.25,'EdgeColor',Color);
+end
+Lg = legend(box_vars([1,4]), {'ikd-Tree','ikd-Forest'},'Location',[0.6707 0.4305 0.265 0.07891],'fontsize',14,'fontname','Times New Roman');
+grid on
+set(gca,'YMinorGrid','off')
+nexttile;
+hold on;
+grid on;
+box on;
+set(gca,'FontSize',12,'FontName','Times New Roman') 
+plot(timestamp_ikd, alpha_bal_ikd,'-','Color',Color_blue,'LineWidth',1.2);
+plot(timestamp_ikd, alpha_del_ikd,'--','Color',Color_orange, 'LineWidth', 1.2);
+plot(timestamp_ikd, 0.6*ones(size(alpha_bal_ikd)), ':','Color','black','LineWidth',1.2);
+lg = legend("\alpha_{bal}", "\alpha_{del}",'location',[0.7871 0.1131 0.1433 0.069],'fontsize',14,'fontname','Times New Roman')
+title("Re-balancing Criterion",'FontSize',16,'FontName','Times New Roman')
+xlabel("time/s",'FontSize',16,'FontName','Times New Roman')
+yl = ylabel("\alpha",'FontSize',15, 'Position',[285.7 0.4250 -1])
+xlim([32,390]);
+ylim([0,0.85]);
+ax3 = gca;
+ax3.YAxis.FontSize = 12;
+ax3.XAxis.FontSize = 12;
+% print('./Figures/fastlio_exp_combine','-depsc','-r1200')
+% exportgraphics(f,'./Figures/fastlio_exp_combine_1.pdf','ContentType','vector')
+

FAST-LIO/Log/guide.md

@@ -0,0 +1 @@
+Here saved the debug records which can be drew by the ../Log/plot.py. The record function can be found frm the MACRO: DEBUG_FILE_DIR(name) in common_lib.h.

FAST-LIO/Log/plot.py

@@ -0,0 +1,94 @@
+# import matplotlib
+# matplotlib.use('Agg')
+import numpy as np
+import matplotlib.pyplot as plt
+
+
+#######for ikfom
+fig, axs = plt.subplots(4,2)
+lab_pre = ['', 'pre-x', 'pre-y', 'pre-z']
+lab_out = ['', 'out-x', 'out-y', 'out-z']
+plot_ind = range(7,10)
+a_pre=np.loadtxt('mat_pre.txt')
+a_out=np.loadtxt('mat_out.txt')
+time=a_pre[:,0]
+axs[0,0].set_title('Attitude')
+axs[1,0].set_title('Translation')
+axs[2,0].set_title('Extrins-R')
+axs[3,0].set_title('Extrins-T')
+axs[0,1].set_title('Velocity')
+axs[1,1].set_title('bg')
+axs[2,1].set_title('ba')
+axs[3,1].set_title('Gravity')
+for i in range(1,4):
+    for j in range(8):
+        axs[j%4, j/4].plot(time, a_pre[:,i+j*3],'.-', label=lab_pre[i])
+        axs[j%4, j/4].plot(time, a_out[:,i+j*3],'.-', label=lab_out[i])
+for j in range(8):
+    # axs[j].set_xlim(386,389)
+    axs[j%4, j/4].grid()
+    axs[j%4, j/4].legend()
+plt.grid()
+#######for ikfom#######
+
+
+#### Draw IMU data
+fig, axs = plt.subplots(2)
+imu=np.loadtxt('imu.txt')
+time=imu[:,0]
+axs[0].set_title('Gyroscope')
+axs[1].set_title('Accelerameter')
+lab_1 = ['gyr-x', 'gyr-y', 'gyr-z']
+lab_2 = ['acc-x', 'acc-y', 'acc-z']
+for i in range(3):
+    # if i==1:
+    axs[0].plot(time, imu[:,i+1],'.-', label=lab_1[i])
+    axs[1].plot(time, imu[:,i+4],'.-', label=lab_2[i])
+for i in range(2):
+    # axs[i].set_xlim(386,389)
+    axs[i].grid()
+    axs[i].legend()
+plt.grid()
+
+# #### Draw time calculation
+# plt.figure(3)
+# fig = plt.figure()
+# font1 = {'family' : 'Times New Roman',
+# 'weight' : 'normal',
+# 'size'   : 12,
+# }
+# c="red"
+# a_out1=np.loadtxt('Log/mat_out_time_indoor1.txt')
+# a_out2=np.loadtxt('Log/mat_out_time_indoor2.txt')
+# a_out3=np.loadtxt('Log/mat_out_time_outdoor.txt')
+# # n = a_out[:,1].size
+# # time_mean = a_out[:,1].mean()
+# # time_se   = a_out[:,1].std() / np.sqrt(n)
+# # time_err  = a_out[:,1] - time_mean
+# # feat_mean = a_out[:,2].mean()
+# # feat_err  = a_out[:,2] - feat_mean
+# # feat_se   = a_out[:,2].std() / np.sqrt(n)
+# ax1 = fig.add_subplot(111)
+# ax1.set_ylabel('Effective Feature Numbers',font1)
+# ax1.boxplot(a_out1[:,2], showfliers=False, positions=[0.9])
+# ax1.boxplot(a_out2[:,2], showfliers=False, positions=[1.9])
+# ax1.boxplot(a_out3[:,2], showfliers=False, positions=[2.9])
+# ax1.set_ylim([0, 3000])
+
+# ax2 = ax1.twinx()
+# ax2.spines['right'].set_color('red')
+# ax2.set_ylabel('Compute Time (ms)',font1)
+# ax2.yaxis.label.set_color('red')
+# ax2.tick_params(axis='y', colors='red')
+# ax2.boxplot(a_out1[:,1]*1000, showfliers=False, positions=[1.1],boxprops=dict(color=c),capprops=dict(color=c),whiskerprops=dict(color=c))
+# ax2.boxplot(a_out2[:,1]*1000, showfliers=False, positions=[2.1],boxprops=dict(color=c),capprops=dict(color=c),whiskerprops=dict(color=c))
+# ax2.boxplot(a_out3[:,1]*1000, showfliers=False, positions=[3.1],boxprops=dict(color=c),capprops=dict(color=c),whiskerprops=dict(color=c))
+# ax2.set_xlim([0.5, 3.5])
+# ax2.set_ylim([0, 100])
+
+# plt.xticks([1,2,3], ('Outdoor Scene', 'Indoor Scene 1', 'Indoor Scene 2'))
+# # # print(time_se)
+# # # print(a_out3[:,2])
+# plt.grid()
+# plt.savefig("time.pdf", dpi=1200)
+plt.show()

FAST-LIO/PCD/1

@@ -0,0 +1 @@
+1