From 0 to 1: Defining the Product Design Architecture for UCOMP AET Remote O&M Platform

Project Background

As UISEE's autonomous driving business transitioned from technology validation to large-scale commercial deployment, front-line operations faced a fundamental leap — from managing individual test vehicles to orchestrating fleets of hundreds of driverless transport vehicles. In the early stages, the absence of a unified centralized management tool meant operations remained heavily reliant on on-site manual intervention and fragmented raw logs. The internal platform operations team identified this critical gap and initiated a product requirement: to build the UISEE Central Remote Operations Platform — a system capable of carrying real-time monitoring responsibilities for all autonomous transport vehicles across every active project.

As the lead designer on this project, I took ownership of the end-to-end design of UCOMP, from zero to one. The goal was to create a "cloud command center" capable of supporting 400+ vehicle fleets at scale, integrating real-time situational awareness with remote safe takeover capabilities — filling the company's gap in large-scale unmanned fleet remote operations infrastructure. We set out to build a centralized "digital brain" that could translate complex vehicle-layer signals into intuitive operational language, ensuring that platform operators could maintain precise command over every vehicle's status, even in demanding industrial environments.

Core Challenge

To break free from the constraints of "closed-loop development", I traveled from UISEE's Shanghai R&D center to the Wuhan Operations Center in 2023 for an in-depth field research visit. Through intensive interviews with 20 operations staff working rotating three-shift schedules — monitoring multi-screen stations around the clock — I discovered that the real challenges were far more complex than anticipated. Under the existing "single-vehicle surveillance" model, operators were prone to fatigue and cognitive overload, with significant delays in detecting anomalies. The key challenges identified at the outset were:

Defining an uncharted operations paradigm：With no established reference model, how do we define a standardized interaction workflow for operators — covering everything from anomaly detection to remote intervention?

Visual re-encoding of extreme data dimensionality：With 400+ mobile endpoints streaming terabytes of data in real time (video feeds, sensor telemetry, path information), how do we perform lossless dimensionality reduction on a single platform and translate it into intuitive visual feedback?

Cross-disciplinary collaboration boundaries：As a ground-up project, we needed to deeply align the underlying algorithmic logic of the autonomous vehicle brain with the performance limits of the frontend display — finding the optimal balance between technical feasibility and peak user experience.

Design Goals

Following a series of interviews and in-depth reflection, we defined the design goals for this new central operations platform as follows:

User Personas

Drawing from in-depth interviews with 20 front-line operators at the Wuhan Operations Center, I built the core user model. These are not merely the audience for visual design — they are the definers of system logic.

Core User · Persona 01

Wuhan Operations Center, 2023

Ops Specialist

400+ AETs 24 / 7 3-shift rotation

Work Background

With foundational knowledge in autonomous driving technology or logistics dispatch, responsible for monitoring the operational status of 400+ AETs nationwide.

Core Tasks

Daily monitoring (keeping vehicles online)

Anomaly handling (responding to alerts)

Remote takeover (long-tail scenarios)

Shift handover records

Behavioral Traits

Rapid Switching Eyes constantly shift between 3–4 screens, processing fragmented information in parallel.

Physical Challenge 3-shift rotation (24/7); attention fluctuates significantly during overnight hours, sensitive to bright light and high-frequency noise.

Key Pain Points

Finding anomalies within massive data streams takes too long

Perceptual lag of the vehicle's surrounding environment during remote takeover

Scenario Simulation

To ensure full-scenario coverage from zero to one, I consolidated the fragmented situations captured during field research into three standardized business scenarios and conducted an interaction walkthrough for each.

Competitive / Reference Research

We analyzed leading design systems including Material Design, Atlassian Design System, IBM Carbon, and Shopify Polaris. This research helped us understand best practices in component architecture, documentation structure, accessibility implementation, and design token management.

Design System Architecture

Facing real-time dynamic data continuously streamed back from 400+ autonomous vehicles, the sheer volume of data made tiered management of such a large fleet the primary challenge I had to address. The greatest design challenge lay in balancing comprehensive monitoring with focused operation. To this end, I built a "funnel-style" information filtering architecture — through view-switching across three dimensions, operators can maintain both a global overview and micro-level focus, enabling meaningful conversion of massive data into high-value decision information:

Design Tokens

We established a comprehensive design token system that serves as the single source of truth for all design decisions. Tokens cover colors, typography, spacing, borders, shadows, and animation timings. This approach ensures consistency across platforms and makes theme customization straightforward.

Visualization Map Design

Anomaly Alert Grading System

如何让运维人员在 1 秒内发现最紧急的问题是在用户调研中常常提及的问题，因此将无人车所涉及到上百种告警内容，我们需要进行分级响应。

Remote O&M Suite

从发现异常到手动远程干预的流程，需要在无人车单车运维界面完成最后的闭环路径。

品牌语言

品牌命名源自 UISEE Central Operation and Maintenance Platform 的首字母缩写，传递出系统化、专业化的产品定位。品牌标识以象征「U」的建筑结构图形为原型，结合主题绿（#00A38F）与深色背景，构建清晰有力的视觉识别。整体品牌语气严谨高效，面向专业运维人员，摒弃一切冗余的装饰性元素。

视觉体系

视觉体系以深色优先为核心原则，降低长班运维中的视觉疲劳。色彩系统结构化地传递运营状态：主题绿（#00A38F）代表正常，琥珀色代表预警，红色代表紧急告警。字体层级经过严格设计，针对高密度数据场景保证可读性，配合统一的间距与网格系统，使界面在信息密度极高的情况下仍能保持视觉秩序。

高保真功能交互

高保真原型覆盖了完整的运维闭环——从车队概览到单车远程干预。关键交互包括：具有分级视听反馈的告警通知系统、带权限状态流转的一键接管流程（空闲 → 申请中 → 操控中 → 释放），以及用于远程驾驶参考的 3D 激光雷达与摄像头融合视图。所有交互均基于真实运维场景验证，以确保决策速度和防误操作能力。

1 / 3

成果与影响

实际落地及规模化运营

UCOMP 已在驭势科技多个商业落地项目中稳定运营，覆盖国内外多个物流园区与无人配送场景。平台实时监管超过 400+ 辆在线无人车，支持 20+ 运维人员并行作业，系统日均处理告警事件数千条。规模化部署验证了 UCOMP 在高并发、复杂场景下的可靠性与稳定性。

行业认可与荣誉

UCOMP 的设计成果获得了行业的广泛关注与认可。项目获得多个设计奖项提名，并在自动驾驶运营领域的行业会议中被作为典型案例分享。UCOMP 的设计方法论也为公司后续产品的设计规范建立提供了重要参考。

经验总结