Understanding the Agility of Running Birds: Sensorimotor and Mechanical Factors in Avian Bipedal Locomotion

Abstract

Birds are a diverse and agile lineage of vertebrates that all use bipedal locomotion for at least part of their life. Thus birds provide a valuable opportunity to investigate how biomechanics and sensorimotor control are integrated for agile bipedal locomotion. This review summarizes recent work using terrain perturbations to reveal neuromechanical control strategies used by ground birds to achieve robust, stable, and agile running. Early experiments in running guinea fowl aimed to reveal the immediate intrinsic mechanical response to an unexpected drop ("pothole") in terrain. When navigating the pothole, guinea fowl experience large changes in leg posture in the perturbed step, which correlates strongly with leg loading and perturbation recovery. Analysis of simple theoretical models of running has further confirmed the crucial role of swing-leg trajectory control for regulating foot contact timing and leg loading in uneven terrain. Coupling between body and leg dynamics results in an inherent trade-off in swing leg retraction rate for fall avoidance versus injury avoidance. Fast leg retraction minimizes injury risk, but slow leg retraction minimizes fall risk. Subsequent experiments have investigated how birds optimize their control strategies depending on the type of perturbation (pothole, step, obstacle), visibility of terrain, and with ample practice negotiating terrain features. Birds use several control strategies consistently across terrain contexts: (1) independent control of leg angular cycling and leg length actuation, which facilitates dynamic stability through simple control mechanisms, (2) feedforward regulation of leg cycling rate, which tunes foot-contact timing to maintain consistent leg loading in uneven terrain (minimizing fall and injury risks), (3) load-dependent muscle actuation, which rapidly adjusts stance push-off and stabilizes body mechanical energy, and (4) multi-step recovery strategies that allow body dynamics to transiently vary while tightly regulating leg loading to minimize risks of fall and injury. In future work, it will be interesting to investigate the learning and adaptation processes that allow animals to adjust neuromechanical control mechanisms over short and long timescales.

Fig. 1.

Schematic illustration of the hierarchical organization of vertebrate neuromechanical control. Transmission delays lead to a temporal scaling of sensorimotor processes that relate to the anatomical distances between sensors, neural networks and effectors. Consequently, central, peripheral and mechanical mechanisms must be integrated over short and long timescales. The fastest responses occur in the periphery, through intrinsic mechanics, intermediate responses occur through short-latency spinal reflexes, and slower responses involve processing and planning in higher brain centers.

Fig. 2.

A trade-off in control of leg retraction rate for terrain robustness versus injury avoidance, illustrated by two boundary conditions. (A) Running dynamics modeled as a SLIP with the swing leg retracted toward the ground just before stance. Leg retraction rate influences the mechanical response in uneven terrain: (B) Fast leg retraction results in steeper leg contact angles and minimizes fluctuations in leg loading, but if leg loading angle (β_TD) reaches 90-degrees, the leg will miss stance, risking a fall. Maximum terrain drop before a fall decreases with increasing rate of leg retraction. (C) Slow leg retraction ensures leg contact, minimizing fall risk, but incurs higher fluctuations in leg loading. Evidence suggests that birds optimize their leg retraction rate to minimize fluctuations in leg-loading (Blum et al. 2014), using intermediate leg retraction rates that ensure contact while avoiding overload injury.

Fig. 3.

Leg length and leg angular trajectories of pheasants negotiating visible obstacles, illustrating a typical three-step strategy. At top, schematic illustration of the landing and take-off conditions of the bird during the step preceding (Step −1), the step on the obstacle (Step 0), and the obstacle dismount (Step +1). Below, leg trajectory (length and angle) during running on level terrain (thin black lines, mean and 95% confidence intervals) and over an obstacle height of 30% leg length (thicker gray lines). Upward triangles indicate foot take-off at the end of stance. Leg length exhibits high stride-to-stride variance in uneven terrain, whereas leg angular trajectory follows a relatively consistent sinusoidal trajectory, with only subtle changes in rate in anticipation of terrain height changes. Data from Birn-Jeffery and Daley (2012).

Fig. 4.

LG muscle length, force, and activation during the immediate response to a hidden pothole perturbation. Figure modified from Daley et al. (2009). At top, the guinea fowl is pictured at the time of ground contact after breaking through the false-floor of tissue paper. Below, thin lines indicate the mean and 95% confidence intervals for steady level running, and thick lines illustrate a perturbed drop step. Force and length are rapidly altered in response to the perturbation, although muscle activation (EMG) remains similar to the level terrain condition.

Fig. 5.

Load- and posture-dependent actuation of the LG muscle during negotiation of uneven terrain. When leg posture is altered at the time of foot contact, altering the balance between muscle and external forces, muscle length during force development (L_t50) varies. L_t50 is the largest predictor of the force and total work output of the muscle (W_net) ( Daley et al. 2009, Daley and Biewener 2011 ). This posture-dependent response is similar between unexpected perturbations and repeating obstacles. This suggests similar task-level control strategies across context, despite potential for differing contributions of intrinsic mechanical, feedforward, and feedback control mechanisms to the response.

Fig. 6.

Context dependent-shifts in the contribution of predictive and reactive modulation of LG activity during obstacle negotiation. Guinea fowl running over obstacles on a treadmill encountered a single footfall on an obstacle (black box) approximately once in 5–7 steps. Step ID corresponds to the sequence of steps with the obstacle encounter at step zero. LG exhibits predictive increases in muscle activity at slower walking speeds (0.7 m/s). Predictive shifts are larger when the obstacles are more visible (higher contrast) relative to the level terrain. At higher speeds (1.3 m/s) guinea fowl use a reactive strategy, with increases in LG activity occurring after foot contact with the obstacle. The influence of high versus low contrast terrain is greater at slower speeds, when the bird has a longer time to process visual information to modulate muscle activity.