A system hierarchy for brain-inspired computing.

Neuromorphic computing draws inspiration from the brain to provide computing technology and architecture with the potential to drive the next wave of computer engineering1-13. Such brain-inspired computing also provides a promising platform for the development of artificial general intelligence14,15. However, unlike conventional computing systems, which have a well established computer hierarchy built around the concept of Turing completeness and the von Neumann architecture16-18, there is currently no generalized system hierarchy or understanding of completeness for brain-inspired computing. This affects the compatibility between software and hardware, impairing the programming flexibility and development productivity of brain-inspired computing. Here we propose 'neuromorphic completeness', which relaxes the requirement for hardware completeness, and a corresponding system hierarchy, which consists of a Turing-complete software-abstraction model and a versatile abstract neuromorphic architecture. Using this hierarchy, various programs can be described as uniform representations and transformed into the equivalent executable on any neuromorphic complete hardware-that is, it ensures programming-language portability, hardware completeness and compilation feasibility. We implement toolchain software to support the execution of different types of program on various typical hardware platforms, demonstrating the advantage of our system hierarchy, including a new system-design dimension introduced by the neuromorphic completeness. We expect that our study will enable efficient and compatible progress in all aspects of brain-inspired computing systems, facilitating the development of various applications, including artificial general intelligence.

Towards artificial general intelligence with hybrid Tianjic chip architecture.

There are two general approaches to developing artificial general intelligence (AGI)1: computer-science-oriented and neuroscience-oriented. Because of the fundamental differences in their formulations and coding schemes, these two approaches rely on distinct and incompatible platforms2-8, retarding the development of AGI. A general platform that could support the prevailing computer-science-based artificial neural networks as well as neuroscience-inspired models and algorithms is highly desirable. Here we present the Tianjic chip, which integrates the two approaches to provide a hybrid, synergistic platform. The Tianjic chip adopts a many-core architecture, reconfigurable building blocks and a streamlined dataflow with hybrid coding schemes, and can not only accommodate computer-science-based machine-learning algorithms, but also easily implement brain-inspired circuits and several coding schemes. Using just one chip, we demonstrate the simultaneous processing of versatile algorithms and models in an unmanned bicycle system, realizing real-time object detection, tracking, voice control, obstacle avoidance and balance control. Our study is expected to stimulate AGI development by paving the way to more generalized hardware platforms.

Learning Quadrupedal High-Speed Running on Uneven Terrain.

Reinforcement learning (RL)-based controllers have been applied to the high-speed movement of quadruped robots on uneven terrains. The external disturbances increase as the robot moves faster on such terrains, affecting the stability of the robot. Many existing RL-based methods adopt higher control frequencies to respond quickly to the disturbance, which requires a significant computational cost. We propose a control framework that consists of an RL-based control policy updating at a low frequency and a model-based joint controller updating at a high frequency. Unlike previous methods, our policy outputs the control law for each joint, executed by the corresponding high-frequency joint controller to reduce the impact of external disturbances on the robot. We evaluated our method on various simulated terrains with height differences of up to 6 cm. We achieved a running motion of 1.8 m/s in the simulation using the Unitree A1 quadruped. The RL-based control policy updates at 50 Hz with a latency of 20 ms, while the model-based joint controller runs at 1000 Hz. The experimental results show that the proposed framework can overcome the latency caused by low-frequency updates, making it applicable for real-robot deployment.

Special Issue: Design and Control of a Bio-Inspired Robot.

Bionics, the interdisciplinary field that draws inspiration from nature to design and develop innovative technologies, has paved the way for the creation of "bio-inspired robots" [...].

CBMC: A Biomimetic Approach for Control of a 7-Degree of Freedom Robotic Arm.

Many approaches inspired by brain science have been proposed for robotic control, specifically targeting situations where knowledge of the dynamic model is unavailable. This is crucial because dynamic model inaccuracies and variations can occur during the robot's operation. In this paper, inspired by the central nervous system (CNS), we present a CNS-based Biomimetic Motor Control (CBMC) approach consisting of four modules. The first module consists of a cerebellum-like spiking neural network that employs spiking timing-dependent plasticity to learn the dynamics mechanisms and adjust the synapses connecting the spiking neurons. The second module constructed using an artificial neural network, mimicking the regulation ability of the cerebral cortex to the cerebellum in the CNS, learns by reinforcement learning to supervise the cerebellum module with instructive input. The third and last modules are the cerebral sensory module and the spinal cord module, which deal with sensory input and provide modulation to torque commands, respectively. To validate our method, CBMC was applied to the trajectory tracking control of a 7-DoF robotic arm in simulation. Finally, experiments are conducted on the robotic arm using various payloads, and the results of these experiments clearly demonstrate the effectiveness of the proposed methodology.

Dynamic Balancing of Humanoid Robot with Proprioceptive Actuation: Systematic Design of Algorithm, Software, and Hardware.

For humanoid robots, maintaining a dynamic balance against uncertain disturbance is crucial, and this function can be achieved by coordinating the whole body to perform multiple tasks simultaneously. Researchers generally accept hierarchical whole-body control (WBC) to address this function. Although experts can build feasible hierarchies using prior knowledge, real-time WBC is still challenging because it often requires a quadratic program with multiple inequality constraints. In addition, the torque tracking performance of the WBC algorithm will be affected by uncertain factors such as joint friction for a large transmission ratio proprioceptive-actuated robot. Therefore, the balance control of physical robots requires a systematic solution. In this study, a robot control system with high computing power and real-time communication ability, UBTMaster, is implemented to achieve a reduced WBC in real time. Based on these, a whole-body control scheme based on task priority for the dynamic balance of humanoid robots is implemented. After realizing the joint friction model identification, finally, a variety of balancing scenarios are tested on the Walker3 humanoid robot driven by the proprioceptive actuators to verify the effectiveness of the proposed scheme. The Walker3 robot exhibits excellent balance when multiple external disturbances occur simultaneously. For example, the two feet of the robot are subjected to tilt and displacement perturbations, respectively, while the torso is subjected to external shocks simultaneously. The experimental results show that the dynamic balance of the robot under multiple external disturbances can be achieved by using strictly hierarchical real-time WBC with a systematic design.

Balanced Standing on One Foot of Biped Robot Based on Three-Particle Model Predictive Control.

Balancing is a fundamental task in the motion control of bipedal robots. Compared to two-foot balancing, one-foot balancing introduces new challenges, such as a smaller supporting polygon and control difficulty coming from the kinematic coupling between the center of mass (CoM) and the swinging leg. Although nonlinear model predictive control (NMPC) may solve this problem, it is not feasible to implement it on the actual robot because of its large amount of calculation. This paper proposes the three-particle model predictive control (TP-MPC) approach. It combines with the hierarchical whole-body control (WBC) to solve the one-leg balancing problem in real time. The bipedal robot's torso and two legs are modeled as three separate particles without inertia. The TP-MPC generates feasible swing leg trajectories, followed by the WBC to adjust the robot's center of mass. Since the three-particle model is linear, the TP-MPC requires less computational cost, which implies real-time execution on an actual robot. The proposed method is verified in simulation. Simulation results show that our method can resist much larger external disturbance than the WBC-only control scheme.

Brain-inspired global-local learning incorporated with neuromorphic computing.

There are two principle approaches for learning in artificial intelligence: error-driven global learning and neuroscience-oriented local learning. Integrating them into one network may provide complementary learning capabilities for versatile learning scenarios. At the same time, neuromorphic computing holds great promise, but still needs plenty of useful algorithms and algorithm-hardware co-designs to fully exploit its advantages. Here, we present a neuromorphic global-local synergic learning model by introducing a brain-inspired meta-learning paradigm and a differentiable spiking model incorporating neuronal dynamics and synaptic plasticity. It can meta-learn local plasticity and receive top-down supervision information for multiscale learning. We demonstrate the advantages of this model in multiple different tasks, including few-shot learning, continual learning, and fault-tolerance learning in neuromorphic vision sensors. It achieves significantly higher performance than single-learning methods. We further implement the model in the Tianjic neuromorphic platform by exploiting algorithm-hardware co-designs and prove that the model can fully utilize neuromorphic many-core architecture to develop hybrid computation paradigm.

A framework for the general design and computation of hybrid neural networks.

There is a growing trend to design hybrid neural networks (HNNs) by combining spiking neural networks and artificial neural networks to leverage the strengths of both. Here, we propose a framework for general design and computation of HNNs by introducing hybrid units (HUs) as a linkage interface. The framework not only integrates key features of these computing paradigms but also decouples them to improve flexibility and efficiency. HUs are designable and learnable to promote transmission and modulation of hybrid information flows in HNNs. Through three cases, we demonstrate that the framework can facilitate hybrid model design. The hybrid sensing network implements multi-pathway sensing, achieving high tracking accuracy and energy efficiency. The hybrid modulation network implements hierarchical information abstraction, enabling meta-continual learning of multiple tasks. The hybrid reasoning network performs multimodal reasoning in an interpretable, robust and parallel manner. This study advances cross-paradigm modeling for a broad range of intelligent tasks.

A hybrid and scalable brain-inspired robotic platform.

Recent years have witnessed tremendous progress of intelligent robots brought about by mimicking human intelligence. However, current robots are still far from being able to handle multiple tasks in a dynamic environment as efficiently as humans. To cope with complexity and variability, further progress toward scalability and adaptability are essential for intelligent robots. Here, we report a brain-inspired robotic platform implemented by an unmanned bicycle that exhibits scalability of network scale, quantity and diversity to handle the changing needs of different scenarios. The platform adopts rich coding schemes and a trainable and scalable neural state machine, enabling flexible cooperation of hybrid networks. In addition, an embedded system is developed using a cross-paradigm neuromorphic chip to facilitate the implementation of diverse neural networks in spike or non-spike form. The platform achieved various real-time tasks concurrently in different real-world scenarios, providing a new pathway to enhance robots' intelligence.