Milad Shafiee

Milad Shafiee, Guillaume Bellegarda, Auke Ijspeert

Learning a locomotion policy for quadruped robots has traditionally been constrained to a specific robot morphology, mass, and size. The learning process must usually be repeated for every new robot, where hyperparameters and reward function weights must be re-tuned to maximize performance for each new system. Alternatively, attempting to train a single policy to accommodate different robot sizes, while maintaining the same degrees of freedom (DoF) and morphology, requires either complex learning frameworks, or mass, inertia, and dimension randomization, which leads to prolonged training periods. In our study, we show that drawing inspiration from animal motor control allows us to effectively train a single locomotion policy capable of controlling a diverse range of quadruped robots. The robot differences encompass: a variable number of DoFs, (i.e. 12 or 16 joints), three distinct morphologies, a broad mass range spanning from 2 kg to 200 kg, and nominal standing heights ranging from 18 cm to 100 cm. Our policy modulates a representation of the Central Pattern Generator (CPG) in the spinal cord, effectively coordinating both frequencies and amplitudes of the CPG to produce rhythmic output (Rhythm Generation), which is then mapped to a Pattern Formation (PF) layer. Across different robots, the only varying component is the PF layer, which adjusts the scaling parameters for the stride height and length. Subsequently, we evaluate the sim-to-real transfer by testing the single policy on both the Unitree Go1 and A1 robots. Remarkably, we observe robust performance, even when adding a 15 kg load, equivalent to 125% of the A1 robot's nominal mass.

Wanming Yu, Chuanyu Yang, Christopher McGreavy et al.

Robot motor skills can be learned through deep reinforcement learning (DRL) by neural networks as state-action mappings. While the selection of state observations is crucial, there has been a lack of quantitative analysis to date. Here, we present a systematic saliency analysis that quantitatively evaluates the relative importance of different feedback states for motor skills learned through DRL. Our approach can identify the most essential feedback states for locomotion skills, including balance recovery, trotting, bounding, pacing and galloping. By using only key states including joint positions, gravity vector, base linear and angular velocities, we demonstrate that a simulated quadruped robot can achieve robust performance in various test scenarios across these distinct skills. The benchmarks using task performance metrics show that locomotion skills learned with key states can achieve comparable performance to those with all states, and the task performance or learning success rate will drop significantly if key states are missing. This work provides quantitative insights into the relationship between state observations and specific types of motor skills, serving as a guideline for robot motor learning. The proposed method is applicable to differentiable state-action mapping, such as neural network based control policies, enabling the learning of a wide range of motor skills with minimal sensing dependencies.

Milad Shafiee, Guillaume Bellegarda, Auke Ijspeert

Quadruped animal locomotion emerges from the interactions between the spinal central pattern generator (CPG), sensory feedback, and supraspinal drive signals from the brain. Computational models of CPGs have been widely used for investigating the spinal cord contribution to animal locomotion control in computational neuroscience and in bio-inspired robotics. However, the contribution of supraspinal drive to anticipatory behavior, i.e. motor behavior that involves planning ahead of time (e.g. of footstep placements), is not yet properly understood. In particular, it is not clear whether the brain modulates CPG activity and/or directly modulates muscle activity (hence bypassing the CPG) for accurate foot placements. In this paper, we investigate the interaction of supraspinal drive and a CPG in an anticipatory locomotion scenario that involves stepping over gaps. By employing deep reinforcement learning (DRL), we train a neural network policy that replicates the supraspinal drive behavior. This policy can either modulate the CPG dynamics, or directly change actuation signals to bypass the CPG dynamics. Our results indicate that the direct supraspinal contribution to the actuation signal is a key component for a high gap crossing success rate. However, the CPG dynamics in the spinal cord are beneficial for gait smoothness and energy efficiency. Moreover, our investigation shows that sensing the front feet distances to the gap is the most important and sufficient sensory information for learning gap crossing. Our results support the biological hypothesis that cats and horses mainly control the front legs for obstacle avoidance, and that hind limbs follow an internal memory based on the front limbs' information. Our method enables the quadruped robot to cross gaps of up to 20 cm (50% of body-length) without any explicit dynamics modeling or Model Predictive Control (MPC).

Milad Shafiee, Guillaume Bellegarda, Auke Ijspeert

Quadruped animals seamlessly transition between gaits as they change locomotion speeds. While the most widely accepted explanation for gait transitions is energy efficiency, there is no clear consensus on the determining factor, nor on the potential effects from terrain properties. In this article, we propose that viability, i.e. the avoidance of falls, represents an important criterion for gait transitions. We investigate the emergence of gait transitions through the interaction between supraspinal drive (brain), the central pattern generator in the spinal cord, the body, and exteroceptive sensing by leveraging deep reinforcement learning and robotics tools. Consistent with quadruped animal data, we show that the walk-trot gait transition for quadruped robots on flat terrain improves both viability and energy efficiency. Furthermore, we investigate the effects of discrete terrain (i.e. crossing successive gaps) on imposing gait transitions, and find the emergence of trot-pronk transitions to avoid non-viable states. Compared with other potential criteria such as peak forces and energy efficiency, viability is the only improved factor after gait transitions on both flat and discrete gap terrains, suggesting that viability could be a primary and universal objective of gait transitions, while other criteria are secondary objectives and/or a consequence of viability. Moreover, we deploy our learned controller in sim-to-real hardware experiments and demonstrate state-of-the-art quadruped agility in challenging scenarios, where the Unitree A1 quadruped autonomously transitions gaits between trot and pronk to cross consecutive gaps of up to 30 cm (83.3 % of the body-length) at over 1.3 m/s.

Peizhuo Li, Hongyi Li, Ge Sun et al.

Despite recent advances in learning-based controllers for legged robots, deployments in human-centric environments remain limited by safety concerns. Most of these approaches use position-based control, where policies output target joint angles that must be processed by a low-level controller (e.g., PD or impedance controllers) to compute joint torques. Although impressive results have been achieved in controlled real-world scenarios, these methods often struggle with compliance and adaptability when encountering environments or disturbances unseen during training, potentially resulting in extreme or unsafe behaviors. Inspired by how animals achieve smooth and adaptive movements by controlling muscle extension and contraction, torque-based policies offer a promising alternative by enabling precise and direct control of the actuators in torque space. In principle, this approach facilitates more effective interactions with the environment, resulting in safer and more adaptable behaviors. However, challenges such as a highly nonlinear state space and inefficient exploration during training have hindered their broader adoption. To address these limitations, we propose SATA, a bio-inspired framework that mimics key biomechanical principles and adaptive learning mechanisms observed in animal locomotion. Our approach effectively addresses the inherent challenges of learning torque-based policies by significantly improving early-stage exploration, leading to high-performance final policies. Remarkably, our method achieves zero-shot sim-to-real transfer. Our experimental results indicate that SATA demonstrates remarkable compliance and safety, even in challenging environments such as soft/slippery terrain or narrow passages, and under significant external disturbances, highlighting its potential for practical deployments in human-centric and safety-critical scenarios.

Milad Shafiee-Ashtiani, Aghil Yousefi-Koma, Reihaneh Mirjalili et al.

In this paper, a combination of ankle and hip strategy is used for push recovery of a position-controlled humanoid robot. Ankle strategy and hip strategy are equivalent to Center of Pressure (CoP) and Centroidal Moment Pivot (CMP) regulation respectively. For controlling the CMP and CoP we need a torque-controlled robot, however most of the conventional humanoid robots are position controlled. In this regard, we present an efficient way for implementation of the hip and ankle strategies on a position controlled humanoid robot. We employ a feedback controller to compensate the capture point error. Using our scheme, a simple and practical push recovery controller is designed which can be implemented on the most of the conventional humanoid robots without the need for torque sensors. The effectiveness of the proposed approach is verified through push recovery experiments on SURENA-Mini humanoid robot under severe pushes.

Milad Shafiee-Ashtiani, Aghil Yousefi-Koma, Masoud Shariat-Panahi

In this paper, previous works on the Model Predictive Control (MPC) and the Divergent Component of Motion (DCM) for bipedal walking control are extended. To this end, we employ a single MPC which uses a combination of Center of Pressure (CoP) manipulation, step adjustment, and Centroidal Moment Pivot (CMP) modulation to design a robust walking controller. Furthermore, we exploit the concept of time-varying DCM to generalize our walking controller for walking in uneven surfaces. Using our scheme, a general and robust walking controller is designed which can be implemented on robots with different control authorities, for walking on various environments, e.g. uneven terrains or surfaces with a very limited feasible area for stepping. The effectiveness of the proposed approach is verified through simulations on different scenarios and comparison to the state of the art.

Milad Shafiee-Ashtiani, Aghil Yousefi-Koma, Masoud Shariat-Panahi et al.

The three bio-inspired strategies that have been used for balance recovery of biped robots are the ankle, hip and stepping Strategies. However, there are several cases for a biped robot where stepping is not possible, e. g. when the available contact surfaces are limited. In this situation, the balance recovery by modulating the angular momentum of the upper body (Hip-strategy) or the Zero Moment Point (ZMP) (Ankle strategy) is essential. In this paper, a single Model Predictive Control (MPC) scheme is employed for controlling the Capture Point (CP) to a desired position by modulating both the ZMP and the Centroidal Moment Pivot (CMP). The goal of the proposed controller is to control the CP, employing the CMP when the CP is out of the support polygon, and/or the ZMP when the CP is inside the support polygon. The proposed algorithm is implemented on an abstract model of the SURENA III humanoid robot. Obtained results show the effectiveness of the proposed approach in the presence of severe pushes, even when the support polygon is shrunken to a point or a line.

Milad Shafiee-Ashtiani, Aghil Yousefi-Koma, Sahba Iravanimanesh et al.

The direct kinematics analysis is the foundation of implementation of real world application of parallel manipulators. For most parallel manipulators the direct kinematics is challenging. In this paper, for the first time a fast and efficient Homotopy Continuation Method, called the Ostrowski Homotopy continuation method has been implemented to solve the direct and inverse kinematics problem of the parallel manipulators. This method has advantage over conventional numerical iteration methods, which is not rely on the initial values and is more efficient than other continuation method and it can find all solutions of equations without divergence just by changing auxiliary Homotopy function. Numerical example and simulation was done to solve the direct kinematic problem of the 3-UPU parallel manipulator that leads to 16 real solutions. Results obviously reveal the fastness and effectiveness of this method than the conventional Homotopy continuation methods such as Newton Homotopy. The results shows that the Ostrowski-Homotopy reduces computation time up to 80-97 % with more accuracy in solutions in comparison with the Newton Homotopy.