Sensors now show where they are in the viewport, and the Robot panel shows what they measure. 视口里能看到传感器在哪里，Robot 面板里能看到它们测到了什么。
Before · 改前
After · 改后
After: camera frustums, the LiDAR head and the IMU axes are drawn where the sensors are.
Before · 改前
After · 改后
After: live sensor rows, LiDAR colouring with a legend, IMU charts and the camera preview.
The scan with a distance ramp and its range in the legend.
Situation
With IMUs, LiDARs and cameras simulated, the editor had to show them. The first version listed sensors as plain text and drew LiDAR points in eight shades of one colour. You could not see where a camera pointed until you pressed Play. During the session the user asked for better visualization.
Task
Show each sensor where it is, whether it is live, and what it measures, in every viewport mode and in a panel that fits a narrow inspector.
Action
- Sensors in the viewport. Camera frustums with a tab marking image up, LiDAR heads with their field of view, and IMU axes in the viewport's X, Y, Z colours. While simulating, each IMU also draws its measured specific force. These gizmos show while editing too, which makes placing a sensor easy.
- A panel that reads at a glance. One row per sensor with its kind, mount and status (Ready, Live, Note). LiDAR scans can be coloured by intensity, distance, height or class with a Turbo legend. IMUs get four seconds of acceleration and angular velocity per axis. Cameras get a live preview of any output.
- One recording. Record sensors writes the host's IMU and LiDAR data and the camera worker's frames into one folder.
- Fixes found by capturing the real app. The native captures for this post showed three layout problems, each fixed with a regression test. The preset picker had widened the Robot panel from 322 to 402 pt and scrolled it sideways. Stale rows painted "Ready" under "Live". Six camera outputs were elided in a segmented control and now use a full-width picker.
Result
The before/after captures run the same scene. Before, the viewport shows nothing where the sensors are and the panel has no sensor section. After, frustums, the LiDAR head and the IMU axes appear while editing. While playing, the panel shows live rows, the coloured scan with its legend, the IMU charts and the camera. The panel is back to its pre-sensor width.
背景
IMU、LiDAR 和相机都能仿真之后，编辑器得把它们展示出来。第一版用纯文本列出传感器，LiDAR 点只用同一种颜色的八个深浅来画。按下 Play 之前，你看不出相机朝哪儿拍。会话中用户也提出可视化需要更好一些。
目标与难点
在每种视口模式下显示每个传感器的位置、是否在工作、测到了什么；面板要能放进很窄的检查器里。
解决思路与行动
- 视口里画出传感器。 相机画成视锥，用一个小标记指示画面的上方；LiDAR 画出探头和视场；IMU 画出坐标轴，颜色和视口的 X、Y、Z 轴一致。仿真时每个 IMU 还会画出它测到的比力。编辑时这些也会显示，方便摆放传感器。
- 一眼能看懂的面板。 每个传感器一行，显示类型、挂载位置和状态（Ready、Live、Note）。LiDAR 扫描可以按强度、距离、高度或类别上色，并带 Turbo 色带图例。IMU 显示最近 4 秒每个轴的加速度和角速度。相机可以实时预览任意一种输出。
- 一次录制。 Record sensors 把宿主的 IMU、LiDAR 数据和相机 worker 的帧写进同一个文件夹。
- 在真实 app 里截图时发现并修复的问题。 为这篇文章截原生窗口时发现了三个布局问题，都已修复并加了回归测试。一是预设下拉框把 Robot 面板从 322 撑到 402，导致横向滚动。二是旧行没有及时移除，"Ready" 和 "Live" 两个状态标签叠在一起。三是六个相机输出在分段控件里被省略显示，现在改成全宽下拉框。
结果与下一步
改前改后用的是同一个场景。改前，视口里传感器的位置什么都没有，面板也没有传感器区。改后，编辑时就能看到视锥、LiDAR 探头和 IMU 坐标轴。播放时，面板显示实时状态行、上色的点云和图例、IMU 曲线以及相机画面。面板也恢复到了加传感器之前的宽度。
Top comments (0)