| using namespace madrona; | |
| namespace test_utils | |
| { | |
| const float EPSILON = 0.001f; // Decreased epsilon to account for floating point errors. TODO: increase floating point precision and ideally use 1e-6 as epsilon. | |
| std::vector<float> flatten_obs(const py::Tensor &obs); | |
| std::pair<float, float> calcMean(const nlohmann::json &rawJson); | |
| } // namespace utils | |