scatti

KO | EN

gitlite — search

#python #react #ai #typescript #golang #go #android #machine-learning #deep-learning #nodejs #awesome-list #ios #openai #open-source

(★ 11)

Motion Generation for Robots and Machines. Real-time. Jerk-constrained. Time-optimal.

scatti 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# scatti

**English** | [简体中文](README.zh.md)

> **Pronunciation**: Italian /ˈskat.ti/ — `scatti` means "bursts / jolts" (rapid movements), fitting a jerk-limited motion generator.

**Pure C99 port of Ruckig — real-time jerk-limited trajectory generation.**

## Features

- Pure C99, zero third-party dependencies, portable to bare-metal / RT kernels (LinuxCNC RTAPI).
- Arbitrary DOF, position / velocity control, three synchronization modes (time / phase / independent).
- 3rd / 2nd / 1st order models, asymmetric velocity and acceleration limits.
- Full Ruckig Pro feature set: **intermediate waypoints**, **per-section constraints**, **position bounds**, **computation interrupts** (microsecond budgets).
- 3-level CMake optimization switch (`-DSCATTI_OPT_LEVEL={0,1,2}`).
- Optional musl cube root + overridable memory macros for libm-free / RTAPI embedded environments.
- Out-of-the-box Python (cffi) + Rust (bindgen) bindings.

Smooth position + bounded acceleration + piecewise-constant jerk — 3-DOF, 500 Hz real-time generation.

## Performance

7-DOF mean / end-to-end (µs):

| Version | Mean | End-to-end |
|---------|:---:|:---:|
| Ruckig C++ `-O3`                              | 6.65 | 7.55 |
| Ruckig C++ `-O3 -march=native -flto` (fair)   | 6.46 | 7.34 |
| **scatti L2 (libm cbrt)**                     | **6.19** | **7.08** |
| **scatti L2 + fast cbrt** (musl replaces libm) | **6.03** | **6.93** |

scatti L2 is **7%** faster on average than fairly-compiled Ruckig, **~9%** faster than the default Release build.
The built-in musl cbrt adds another 2-3% and removes the libm dependency — suitable for RT kernels.

DOF scaling (fast cbrt mode, same machine):

| DOFs | Mean (µs) | End-to-end (µs) |
|:---:|:---:|:---:|
|  3 | 2.24  | 2.69  |
|  7 | 6.04  | 6.94  |
| 14 | 12.76 | 14.15 |

### Numerical consistency

Cross-checked against C++ Ruckig over 99,344 random trajectories; both libm and fast-cbrt configurations are **bit-identical**:

| Metric | Result |
|--------|--------|
| Match rate | **100.00%** |
| Max duration diff | 4.55e-13 s |
| Max position diff | 5.90e-12 |
| Max velocity diff | 9.24e-14 |

## Quick start

```bash
cmake -B build -DSCATTI_OPT_LEVEL=1
cmake --build build -j
sudo cmake --install build      # or: cd build && cpack   to build a .deb
```

Minimal usage:

```c
#include <scatti/scatti.h>

CRuckig *otg = scatti_create(3, 0.01);       /* 3 DOF, 10 ms cycle */
CRuckigInputParameter *in  = scatti_input_create(3);
CRuckigOutputParameter *out = scatti_output_create(3);

in->current_position[0] = 0.0;  in->target_position[0] = 5.0;
in->max_velocity[0]     = 3.0;
in->max_acceleration[0] = 3.0;
in->max_jerk[0]         = 4.0;
/* fill the remaining DOFs the same way */

while (scatti_update(otg, in, out) == CRuckigWorking) {
    /* drive the actuator: out->new_position / new_velocity / new_acceleration */
    scatti_output_pass_to_input(out, in);
}

scatti_output_destroy(out);
scatti_input_destroy(in);
scatti_destroy(otg);
```

Headers live in `include/scatti/`; the full API and Pro fields are documented in the corresponding header comments.

CMake integration:

```cmake
add_subdirectory(scatti)
target_link_libraries(your_target PRIVATE scatti m)
```

## Acknowledgements

Based on [Ruckig](https://github.com/pantor/ruckig) by Lars Berscheid (original C++ implementation, Community Edition).

Paper: Berscheid, L., & Kroeger, T. (2021). *Jerk-limited Real-time Trajectory Generation with Arbitrary Target States.* Robotics: Science and Systems (RSS). [arXiv:2105.04830](https://arxiv.org/abs/2105.04830)

## License

MIT (same as Ruckig Community Edition).

scatti

scatti 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

scatti

scatti 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README