Learning-Based Model Predictive Control (LMPC) is a class of algorithms that enhances Model Predictive Control (MPC) by including machine learning methods, improving robot navigation in complex environments.However, the combination of machine learning and MPC computation in LMPC creates a unique workload that cannot be efficiently handled by a simple GPU and CPU integration.We present HiPER, a hierarchically-composed processing array that provides temporal and spatial mapping capabilities, allowing efficient adaptation to workload changes at runtime.To simplify control, HiPER employs a pointer queue hierarchy to compose and orchestrate program execution.Additionally, HiPER utilizes a fractal interconnect topology that combines local systolic interconnects and their hierarchical extensions to efficiently support the workload's traffic characteristics.To evaluate the performance and efficiency of HiPER, we synthesized a 16.37 mm 2 design in 16nm CMOS.The design consists of 6 pointer queue levels and 1024 PEs.The prototype was assessed using a representative LMPC workload, demonstrating 10.75× improvement in performance compared to a GTX 1080 GPU, 12.80× improvement in energy efficiency compared to a Jetson Orin Nano embedded GPU, and 11.6×/22.2×improvement in performance compared to the RoboX/Plasticine accelerators.

Neural ordinary differential equations (NODEs) provide better modeling performance with smaller amount of model parameters in many tasks by embedding neural networks (NNs) in ordinary differential equations (ODEs). They have been shown to outperform in representing continuous-time data and learning dynamic systems, and are promising for on-device inference and training. However, an edge device is limited by area and energy budget, and real-time operations have a tight latency requirement. State-of-the-art NN accelerators are not optimized for the area- and power-hungry memory storage and access for NODE inference and training, and lack the flexibility to incorporate dynamic latency reduction techniques. We present eNODE by architecture-algorithm co-design to achieve efficient and fast inference and training of NODEs. eNODE adopts compact-size depth-first integration and depth-first training for higher energy efficiency. Through function reuse, packetized processing and a unified NN core design, the efficiency of eNODE’s depth-first processing is further enhanced. We propose algorithm innovations, including slope-adaptive stepsize search and priority processing with early stop, to substantially shorten the latency. A hardware prototype is synthesized in a 28 nm CMOS technology for evaluation and benchmarking. eNODE demonstrates up to 6.59× better energy efficiency, 2.38× higher speed, and better area scalability over a SIMD ASIC baseline.

Person re-identification, aiming to match a specific person among non-overlapping cameras, has attracted plenty of attention in recent years. It can be regarded as a visual retrieval task, namely given a query person image, ranking all gallery images according to their similarities to the query. Conventionally, this similarity function is learnt by forcing intra-distances to be small while inter-distances to be large, which are referred to as individual similarity constraints. In this paper, we propose to learn the similarity function by taking into account of both individual similarity constraints and contextual similarity constraints. The context of a query is defined as its k-nearest neighbors in the gallery. We argue that if two images are from the same person, apart from the visual likeness between them, denoted as the individual similarity, they should also possess similar k-nearest neighbors in the gallery, denoted as the contextual similarity. Motivated by this assumption, we propose a new Contextual Similarity Regularized Metric Learning (CSRML) method for person re-identification. The contextual similarity regularization term forces two images of the same person to share similar context. Both individual and contextual similarity constraints are encoded by a large margin logistic loss function and the final problem is solved by the stochastic gradient descent algorithm. Experiments on the challenging VIPeR and CUHK01 datasets show that our approach achieves very competitive performance.