(A) Schematic of the obstacle avoidance setup. A head-fixed mouse runs on top of a wheel with a mirror mounted inside, allowing a single camera to capture two orthogonal views. A second camera focuses on the whiskers. An obstacle is moved toward the mouse along a motorized, belt-driven linear track at a speed matched to that of the wheel. (B) DeepLabCut is used to track the positions of the paws, tail, and nose in both views. Tracking from the two views is combined to reconstruct the three-dimensional pose at 250 Hz. (C) Custom convolutional neural network algorithms were developed to determine when the whiskers (bottom row) and paws (top rows) contact the obstacle. (D-J) Behavioral characterization of head-fixed obstacle clearance for n = 20 mice. (D) Obstacle clearance success rate as a function of obstacle height for each paw (mean with S.E.M. shaded). All paws cleared the obstacle at high rates even for high obstacles. (E) Average running velocity as a function of position relative to the obstacle (standard deviation is shaded; positive numbers mean the obstacle is in front of the mouse). Thin lines are averages for individual mice. Vertical black line shows the position at which the obstacle is beneath the nose of the mouse. (F) Example trial showing (from left to right) the leading forelimb (LF), trailing forelimb (TF), leading hindlimb (LH), and trailing hindlimb (TH) clearing the obstacle. Traces show kinematics from trials selected randomly from a single session. (G) Average kinematics across mice for LF, TF, LH, and TH binned by obstacle height (colored traces). Dashed gray traces are the average of the two steps preceding whisker contact with the obstacle. The obstacle is 3.175 mm in diameter. (H) Hildebrand plots (Hildebrand, 1989) , averaged across mice, reveal trot gaits during normal locomotion and obstacle clearance. Color intensity represents likelihood of stance. The top panel is calculated across all steps, and the bottom panel from steps over the obstacle. (I) Paw height vs. obstacle height averaged across mice (S.E.M. is shaded). Height is measured when the paw is 8 mm in front of the obstacle. The dashed gray line is the unity line. (J) Average correlation between the obstacle height and the height of all paws, measured when the paw is 8 mm in front of the obstacle. Circles represent individual mice. For each paw, the correlation is computed for the step over the obstacle (colored circles), and the preceding control steps (dark circles). See also Figure 1—figure supplement 1.