OpenCat robots

Match the actual code to the hardware

The README's setup example selects Bittle and NyBoard V1_0, while the preserved OpenCat.ino actually defaults to Bittle and NyBoard V1_2. The header contains separate board definitions: V1 variants use 115200 baud, while V0_1 and V0_2 use 57600. Copying the README's serial setting without checking the selected board would miss that distinction. [S001] [S005] [S006]

The documented setup has a board-configuration phase followed by the main sketch. IMU calibration and joint-offset calibration are distinct steps. The README says automatic initialization preserves joint offsets, but the pinned implementation disagrees: with AUTO_INIT enabled, src/skill.h removes the confirmation condition and writes the servoCalib array into calibration EEPROM. The IMU code likewise runs its calibration without the prompt in configuration mode. Treat the README’s preservation statement as inconsistent with this release’s code, and do not rely on it to retain existing offsets. No upload, calibration or robot motion was performed in this review. [S001] [S005] [S008] [S009]

The sketch also notes that optional sensor/behavior modes can disable gyro code to conserve program memory. Adding a module therefore has a functional tradeoff, not just a library-installation step. The selected robot, board, modules and skill tables need to be considered together. [S005] [S006]

Available software is not a complete hardware release

The archive includes firmware, module tests, skill libraries, Python API/serial tools and a desktop interface. It also contains two STL artifacts, but their paths identify a servo test arm and a UI tool; they do not establish a complete Nybble or Bittle frame. The README links external build and product resources, which remain separate from this preserved software release. A matched mechanical/electronics manufacturing package and complete BOM have not been established here. [S001] [S007]

The root repository and main sketch declare MIT, while the sketch directs readers to third-party module notices for included code. That software grant is not automatically extended to externally hosted frame designs, board designs or applications. Likewise, links to users' reinforcement-learning demonstrations do not establish bundled, verified learned weights in the reviewed firmware configuration. [S001] [S003] [S005] [S007]

This entry is useful for comparing the two documented NyBoard robot configurations and avoiding setup mismatches. Current full-kit cost, component substitutions and physical performance remain unverified rather than inferred from demonstrations or retail availability claims. [S001] [S005] [S006]