EFRI-M3C: Robust Decoder-Compensator Architecture for Interactive Control of High-Speed and Loaded Movements

目的：假肢的脑-机交互控制（BMIC）是一个重大的挑战，以高速和自然的运动。当前的BMIC范例在大脑和假体之间采用前馈接口，称为“解码器”，其成功在很大程度上依赖于大脑在“特定”环境中适当地利用视觉反馈信息进行适应的能力。这种解码器使用来自健康受试者的数据进行训练，但被实现为脊髓患者的接口。健康受试者的运动皮层输出与受伤患者的运动皮层输出基本上不同，并且解码器不考虑由于本体感受数据的丢失而在小脑中产生的伪信号。 因此，关键的挑战是为未来的BMIC设计鲁棒的解码器，同时考虑小脑和皮层的贡献。 智力优势：我们提出了一种新的鲁棒解码器-补偿器（RDC）架构，用于在存在不确定性的情况下对快速运动进行交互式控制。RDC是一种反馈互连，1）解码皮层信号以产生反映运动意图的致动器命令，2）校正在没有本体感受反馈的情况下产生的虚假小脑信号，以及3）使互连鲁棒以考虑模型和现实之间的失配。从癫痫患者执行快速和负荷运动中获得的运动区的多部位颅内EEG记录将有助于健康和脊髓患者的皮质结构的系统识别。皮层模型和RDC架构将属于一类线性参数变化系统，并且RDC将被合成以在广泛的运动和环境中保持性能。更广泛的影响：我们提供了一个统一的框架，旨在理解人类运动控制，它结合了小脑和皮层模型，并建立了一个快速和自然运动的BMIC，使脊髓损伤和小脑共济失调的患者能够执行快速自然的轨迹。强大的扩展闭环系统识别和鲁棒控制综合技术的LPV系统将被开发。该计划还将为学生提供难得的机会，以解决系统工程和神经生物学之间的接口的挑战。

Objective: Brain-machine interactive control (BMIC) of prosthetic limbs for high speed and natural movements is a major challenge. The current BMIC paradigm employs a feedforward interface between the brain and prosthetic, referred to as the "decoder", whose success relies heavily on the ability of the brain to adapt appropriately utilizing visual feedback information in a "certain" environment. Such decoders are trained using data from healthy subjects but are implemented as interfaces for spinal cord patients. The motor cortical output of the healthy subject is substantially different from that of an injured patient, and decoders do not account for spurious signals generated in the cerebellum due to the loss of proprioceptive data. Thus, the key challenge is to design robust decoders for BMIC of the future that take into account both cerebellar and cortical contributions. Intellectual Merit: We propose a novel Robust Decoder-Compensator (RDC) architecture for interactive control of fast movements in the presence of uncertainty. The RDC is a feedback interconnection that 1) decodes cortical signals to produce actuator commands that reflect motor intent, 2) corrects for spurious cerebellar signals generated in the absence of proprioceptive feedback, and 3) makes robust the interconnection to account for mismatches between models and reality. Multi-site intracranial EEG recordings in motor areas obtained from epilepsy patients executing fast and loaded movements will facilitate system identification of cortical structures in healthy and in spinal cord patients. The cortical models and the RDC architecture will belong to a class of linear parameter varying systems, and the RDC will be synthesized to maintain performance over a wide range of movements and environments. Finally, we will implement the interactive system on patients with implanted electrodes.Broader Impact: We provide a unified framework aimed at understanding human motor control, which incorporates cerebellar and cortical models and builds a BMIC for fast and natural movements, allowing patients with spinal cord injuries and cerebellar ataxia to execute rapid natural trajectories. Powerful extensions of closed-loop system identification and robust control synthesis techniques for LPV systems will be developed. This program will also give students rare opportunities to address challenges at the interface between systems engineering and neurobiology.