feat: support of Hesai FT120 sensor by mgiardino-be · Pull Request #412 · tier4/nebula

mgiardino-be · 2026-02-26T15:16:36Z

PR Type

New Feature

Description

Add support for Hesai FT120 solid state sensor.

Review Procedure

Remarks

Pre-Review Checklist for the PR Author

PR Author should check the checkboxes below when creating the PR.

Assign PR to reviewer

Checklist for the PR Reviewer

Reviewers should check the checkboxes below before approval.

Commits are properly organized and messages are according to the guideline
(Optional) Unit tests have been written for new behavior
PR title describes the changes

Post-Review Checklist for the PR Author

PR Author should check the checkboxes below before merging.

All open points are addressed and tracked via issues or tickets

CI Checks

Build and test for PR: Required to pass before the merge.

…ion binary file. Added calibration file from a sensor.

Added definitions for message tail, data packet and FT120 sensor.

…tatus structure that can be reused (is the same as for AT128 and QT128 sensors). Integrated the sensor in the HW interface.

…tion for each pixel.

- it uses old ScanCutAngles approac (not a big deal due to limited FoV); - dual return works, but manual recommends to use only single return modes; - maybe a couple of "if" in unpack and convert_returns can be simplified; -mask filter was not tested.

Include the new sensor in the driver source, with the correct decoder selection.

mojomex · 2026-03-04T01:27:53Z

@mgiardino-be Thank you for the PR, looks very good already! I will finish my review by end of this week.

In the meantime, could you check and address the failing pre-commit checks? 🙏

mojomex

Hi, sorry for the super long wait 🙇

The code looks pretty good! I've left a few small comments here and there. Since this is the first solid state Hesai LiDAR in Nebula, we'll have to reconsider parameter handling since cloud_{min,max}_angle and a few other params are now obsolete. It's fine as it is in this PR though!

The reason we took so long is that we got word from Hesai that FT120 has been end-of-lifed (there is some public info on the web about this).
Our projects are moving to FTX, and the implementation is different again to the FT120, so we are currently discussing if we still want to merge and maintain your FT120 PR.

We'll let you know ASAP when we have clarity on how to move forward. In the meantime, I want to thank you again for the contribution and apologize that we are only considering this issue after you already went through the trouble of preparing your PR!

FYI @drwnz

mojomex · 2026-02-27T02:16:08Z

src/nebula_hesai/nebula_hesai/config/PandarFT120.param.yaml

+    min_range: 0.3
+    max_range: 22.0
+    cloud_min_angle: 40
+    cloud_max_angle: 139


(unconfirmed) Given angluar azimuth resolution of 0.625deg, rounding down here could cause the last column to be filtered out.

(Nebula would have to be changed to allow centi- or milli-degree settings here to fix this)

I should think how to verify the issue.
Relevant part is at the end of hesai_decoder.hpp.

we received a packet for which passed_timestamp_reset_angle()==true (a new scan)

we should always assign new_scan_timestamp_ns to output_frame_, as we are processing a new scan but the buffers are not swapped yet;

For this reason, we should always impose that cut_angle != cloud_max_angle: due to inner working of the sensor, it made no sense to have a cut_angle which is not associated to the sensor minimum (= transmit old decode frame when we start processing a new one) or maximum column (we have completed one frame, send it). If cut_angle != cloud_max_angle, the timestamp is going to be assigned to output_frame_ and the procedure should be correct.

mojomex · 2026-02-27T02:17:13Z

src/nebula_hesai/nebula_hesai/config/PandarFT120.param.yaml

+    cloud_min_angle: 40
+    cloud_max_angle: 139
+    sync_angle: 40  # 40-139
+    cut_angle: 139.0


(unconfirmed) might cause a column of points from an old scan to turn up in the new scan.

This is linked to above comment.
Worse, there is a bug in the code (at line 575 in hesai_decoder.hpp) so that the first column of a new scan is written into the frame for the old one: I'll apply a bugfix.

mojomex · 2026-02-27T02:21:05Z

src/nebula_hesai/nebula_hesai_common/include/nebula_hesai_common/hesai_common.hpp

+    auto raw_ptr = buf.data();
+    col_count = raw_ptr[6];
+    row_count = raw_ptr[7];
+    resolution = raw_ptr[8];


Raw pointer without bounds checking can cause a segfault or other corruption when an invalidly sized buffer is given. Prefer:

Suggested change

auto raw_ptr = buf.data();

col_count = raw_ptr[6];

row_count = raw_ptr[7];

resolution = raw_ptr[8];

col_count = raw_ptr.at(6);

row_count = raw_ptr.at(7);

resolution = raw_ptr.at(8);

mojomex · 2026-02-27T02:22:39Z

src/nebula_hesai/nebula_hesai_common/include/nebula_hesai_common/hesai_common.hpp

+    const auto count{col_count * row_count};
+    const auto count_bytes{4 * count};
+
+    auto ref = &(buf[9]);


Perform size checking here (are there really count_bytes after the given offset?)

mojomex · 2026-02-27T02:24:07Z

src/nebula_hesai/nebula_hesai_common/include/nebula_hesai_common/hesai_common.hpp

+  [[nodiscard]] std::tuple<float, float> get_fov_padding() const override
+  {
+    // For FT120 should be enough
+    return {-0.1, 0.1};


(note for self) check

mojomex · 2026-02-27T02:26:14Z

src/nebula_hesai/nebula_hesai_common/include/nebula_hesai_common/hesai_common.hpp

    case SensorModel::HESAI_PANDAR128_E3X:
    case SensorModel::HESAI_PANDAR128_E4X:
    case SensorModel::HESAI_PANDARQT128:
+    case SensorModel::HESAI_PANDARFT120:


Same as above

mojomex · 2026-02-27T02:51:27Z

src/nebula_hesai/nebula_hesai_decoders/include/nebula_hesai_decoders/decoders/hesai_packet.hpp

+  uint16_t row_num;     // 120
+  uint8_t column_res;   // 63, to be multiplied by standard coefficient of 0.01°
+  uint8_t row_res;      // 63, to be multiplied by standard coefficient of 0.01°
+  uint8_t return_num;  // 0, single return, 1 dual return & block 1 returns first type of dual mode;


The manual calls this field "first_block_return", might be good to align with that

mojomex · 2026-02-27T02:54:39Z

src/nebula_hesai/nebula_hesai_decoders/include/nebula_hesai_decoders/decoders/pandar_ft120.hpp

+  uint8_t factory_information;  // fixed, 0x42
+  uint32_t udp_sequence;
+  uint32_t crc_tail;
+  uint32_t


uint8_t signature[16] might be more idiomatic since it's just a long number, not a sequence of 4 ints.

mojomex · 2026-03-09T12:27:48Z

...ers/include/nebula_hesai_decoders/decoders/angle_corrector_calibration_based_solid_state.hpp

+
+        ++calib_i;
+
+        auto C = CorrectedAngleData();


Please name more descriptively, e.g. data or corrected_angle_data.

I'm changing it to pixel_angle_data, to avoid problems with corrected_angle_data.

mojomex · 2026-03-09T12:28:14Z

...ers/include/nebula_hesai_decoders/decoders/angle_corrector_calibration_based_solid_state.hpp

+: public AngleCorrector<HesaiSolidStateCalibration, ColumnN>
+{
+private:
+  std::array<std::array<CorrectedAngleData, RowN>, ColumnN> correctedAngleData;


Please name this in snake_case, i.e. corrected_angle_data.

codecov · 2026-03-16T05:10:58Z

Codecov Report

❌ Patch coverage is 0.41152% with 242 lines in your changes missing coverage. Please review.
✅ Project coverage is 35.94%. Comparing base (7a0b309) to head (979f640).
⚠️ Report is 2 commits behind head on main.

❌ Your patch check has failed because the patch coverage (0.00%) is below the target coverage (90.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #412      +/-   ##
==========================================
- Coverage   36.75%   35.94%   -0.81%     
==========================================
  Files         138      140       +2     
  Lines       11207    11461     +254     
  Branches     5960     6047      +87     
==========================================
+ Hits         4119     4120       +1     
- Misses       6301     6549     +248     
- Partials      787      792       +5

Flag	Coverage Δ
nebula	`36.02% <0.41%> (?)`
nebula_core_common	`36.02% <0.41%> (?)`
nebula_hesai	`36.02% <0.41%> (?)`
nebula_hesai_common	`36.02% <0.41%> (?)`
nebula_hesai_decoders	`36.02% <0.41%> (?)`
nebula_hesai_hw_interfaces	`36.02% <0.41%> (?)`

Flags with carried forward coverage won't be shown. Click here to find out more.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

…w scan is received.

mgiardino-be · 2026-03-16T15:58:09Z

Hi, sorry for the super long wait 🙇

The code looks pretty good! I've left a few small comments here and there. Since this is the first solid state Hesai LiDAR in Nebula, we'll have to reconsider parameter handling since cloud_{min,max}_angle and a few other params are now obsolete. It's fine as it is in this PR though!

The reason we took so long is that we got word from Hesai that FT120 has been end-of-lifed (there is some public info on the web about this). Our projects are moving to FTX, and the implementation is different again to the FT120, so we are currently discussing if we still want to merge and maintain your FT120 PR.

We'll let you know ASAP when we have clarity on how to move forward. In the meantime, I want to thank you again for the contribution and apologize that we are only considering this issue after you already went through the trouble of preparing your PR!

FYI @drwnz

I added some commits to apply your suggestions (and, meanwhile thinking about cloud_{min,max}_angle, I should have fixed a bug, in the latest commit). I'm waiting your suggestion for the last lines of hesai_common.hpp, please se my answers above.

For FT120 end-of-life, also our vendor told us the same information, so it is understandable if the PR will not be merged. I'm curious, do you already know how Hesai has changed the packet structure in the new FTX sensor? Maybe it is possible to adapt part of our work (at least, a little hope... 😁).

Have a nice day
MG

mgiardino-be and others added 11 commits February 25, 2026 11:58

Updated launch file and sensor schema.

7361ddf

Added param and schema file for FT120.

dcfba97

Added the sensor in common headers.

a2f1e69

Added HesaiSolidStateCalibration class to read and write the calibrat…

5f33957

…ion binary file. Added calibration file from a sensor.

Added definitions for header, unit and data block.

94b643d

Added definitions for message tail, data packet and FT120 sensor.

Added structures to manage sensor configuration and status. Renamed s…

d7d4d26

…tatus structure that can be reused (is the same as for AT128 and QT128 sensors). Integrated the sensor in the HW interface.

Added the class that computes the lookup tables for azimuth and eleva…

ee41fc2

…tion for each pixel.

Integratedthe new angle corrector into HesaiSensor.

72e5b07

Include the solid state calibration in the ROS wrapper.

30a841c

Include the new sensor in the driver source, with the correct decoder selection.

ci(pre-commit): autofix

94f3387

mojomex requested review from drwnz and mojomex February 27, 2026 02:03

mgiardino-be changed the title ~~Support of Hesai FT120 sensor~~ feat: Support of Hesai FT120 sensor Feb 27, 2026

mgiardino-be changed the title ~~feat: Support of Hesai FT120 sensor~~ feat: support of Hesai FT120 sensor Feb 27, 2026

mgiardino-be and others added 2 commits March 4, 2026 12:16

Add includes as requested by cpplint.

585e85c

ci(pre-commit): autofix

979f640

mojomex reviewed Mar 16, 2026

View reviewed changes

mgiardino-be and others added 7 commits March 16, 2026 12:49

Added safe read and size checking when reading sensor calibration file.

1656be2

Renamed return_num in packet header structure to first_block_return.

7d6338b

Changed AES signature type.

ffd103d

Renamed variables to match coding style and improve

0c9b5d9

ci(pre-commit): autofix

b942dc9

Bugfix: write points into output_frame_ when the first packet of a ne…

49c8f67

…w scan is received.

ci(pre-commit): autofix

77a6e03

Conversation

mgiardino-be commented Feb 26, 2026

PR Type

Related Links

Description

Review Procedure

Remarks

Pre-Review Checklist for the PR Author

Checklist for the PR Reviewer

Post-Review Checklist for the PR Author

CI Checks

Uh oh!

mojomex commented Mar 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mojomex left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mgiardino-be Mar 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

codecov bot commented Mar 16, 2026

Codecov Report

Uh oh!

mgiardino-be commented Mar 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

mojomex commented Mar 4, 2026 •

edited

Loading

mgiardino-be Mar 16, 2026 •

edited

Loading