递归火山软件开发平台

标题: 求助python代码转火山 [打印本页]

作者: 卢本伟 时间: 昨天 16:53
标题: 求助python代码转火山
mport numpy as np
import numpy.typing as npt

class Predictor:
ALPHA_VEL = 0.15
ALPHA_ACC = 0.15

def __init__(self):
      self.smoothed_vel = np.zeros(2)
      self.smoothed_acc = np.zeros(2)
      self.prev_vel = np.zeros(2)

def predict(
      self,
      curr_e: npt.NDArray[np.float64],
      prev_e: npt.NDArray[np.float64],
      prev_m: npt.NDArray[np.float64],
      dt: float,
):
      vel_raw = ((curr_e - prev_e) + prev_m) / dt

      for axis in range(2):
         if np.sign(vel_raw[axis]) != np.sign(curr_e[axis]):
            vel_raw[axis] = 0.0

      self.prev_vel = self.smoothed_vel.copy()
      self.smoothed_vel = (
         self.ALPHA_VEL * vel_raw + (1 - self.ALPHA_VEL) * self.smoothed_vel
      )

      acc_raw = (self.smoothed_vel - self.prev_vel) / dt

      for axis in range(2):
         if np.sign(acc_raw[axis]) != np.sign(curr_e[axis]):
            acc_raw[axis] = 0.0

      self.smoothed_acc = (
         self.ALPHA_ACC * acc_raw + (1 - self.ALPHA_ACC) * self.smoothed_acc
      )

      pred_vel = self.smoothed_vel * dt
      pred_acc = 0.5 * self.smoothed_acc * (dt**2)

      return pred_vel + pred_acc

def reset(self):
      self.smoothed_vel.fill(0)
      self.smoothed_acc.fill(0)
      self.prev_vel.fill(0)
用于物体轨迹追踪

作者: huanhai 时间: 昨天 18:21
(, 下载次数: 8)
这样的吗

作者: a759077146 时间: 昨天 19:11
用卡尔曼预测

作者: 卢本伟 时间: 1 小时前

huanhai 发表于 2025-12-18 18:21
这样的吗

应该是的

欢迎光临递归火山软件开发平台 (https://bbs.voldp.com/)