def reset_state()

in bisk/tasks/gaps.py [0:0]

• 43 lines of code
• 5 McCabe index (conditional complexity)

    def reset_state(self):
        super().reset_state()
        self.max_platforms_reached = 0
        xpos = 4
        if self.fixed_size:
            gaps = np.zeros(self.n_platforms) + self.min_gap
            sizes = np.zeros(self.n_platforms) + self.max_size
        else:
            if self.robot.startswith('quadruped'):
                gaps = self.np_random.uniform(
                    0.8, 1.55, size=(self.n_platforms,)
                )
                ms = max(self.max_size * 2, 2.0)
                sizes = self.np_random.uniform(
                    2.0, ms, size=(self.n_platforms,)
                )
            elif self.robot.startswith('humanoid'):
                gaps = self.np_random.uniform(
                    self.min_gap, self.max_gap, size=(self.n_platforms,)
                )
                sizes = self.np_random.uniform(
                    1.0, self.max_size, size=(self.n_platforms,)
                )
            else:
                gaps = self.np_random.uniform(
                    self.min_gap, self.max_gap, size=(self.n_platforms,)
                )
                sizes = self.np_random.uniform(
                    0.5, self.max_size, size=(self.n_platforms,)
                )

        self.gap_starts = []
        self.platform_starts = []
        for i in range(self.n_platforms):
            self.gap_starts.append(xpos)
            self.p.named.model.geom_size[f'gap-{i}'][0] = gaps[i] / 2
            self.p.named.model.geom_pos[f'gap-{i}'][0] = xpos + gaps[i] / 2
            xpos += gaps[i]
            self.platform_starts.append(xpos)
            self.p.named.model.geom_size[f'platform-{i}'][0] = sizes[i] / 2
            self.p.named.model.geom_pos[f'platform-{i}'][0] = (
                xpos + sizes[i] / 2
            )
            xpos += sizes[i]