Carnegie Mellon Deploys NavLab Truck to Test Onboard Autonomous Drive

NavLab 1 (1986) : Carnegie Mellon : Robotics Institute History of Self-Driving Cars — CMU Robotics Institute

CMU Robotics Institute

Housing computing hardware directly inside the vehicle eliminates telemetry bottlenecks and external infrastructure dependence, allowing a centralized blackboard framework to fuse sensor feeds and control driving.

AI-assisted summary of NavLab 1 (1986) : Carnegie Mellon : Robotics Institute History of Self-Driving Cars — original source. Check the original for full context.

Source published:

Key ideas

  • Carnegie Mellon University's 1986 NavLab 1 pioneered onboard autonomous computing by housing researchers and processing hardware entirely within a modified commercial truck. (0:00)
  • The self-contained mobile architecture bypassed the severe telemetry bottlenecks and communications failures that hindered stationary offboard robotics testbeds. (0:00)
  • NavLab's multimodal perception system integrated color video imagery with scanning laser rangefinder data to construct unified 3D models with geometry and material reflectance. (1:14)
  • A centralized Blackboard software architecture orchestrated the vehicle's asynchronous distributed modules across sensing, mapping, planning, and driving. (1:14)

Detailed notes

## Background Published by the CMU Robotics Institute on YouTube on October 13, 2016, this archival video highlights NavLab 1 from 1986. According to the description, the Navigation Laboratory was the first self-driving vehicle with onboard human passengers, and despite being very slow due to the computing constraints of the era, it laid foundational groundwork for Carnegie Mellon University's autonomous vehicle research.

## TL;DR The video outlines the physical and computational architecture of Carnegie Mellon University's NavLab, an experimental autonomous testbed built inside a modified commercial truck. By housing both the computing hardware and the research team on board, the vehicle avoided early telemetry bottlenecks and allowed researchers to travel directly to remote test sites while retaining manual driving overrides. NavLab fuses multimodal sensory inputs—including color and intensity video cameras alongside an active scanning laser rangefinder—to create detailed natural scene descriptions. These perceptual feeds, mapping systems, path planners, and driving routines are integrated through a centralized Blackboard computer architecture that coordinates the entire autonomous driving pipeline.

## Atmosphere & style The presentation carries the measured, didactic tone of a vintage 1980s corporate and academic research documentary, guided by a steady, formal narrator. It feels earnest and historically pioneering, presenting an inside look at what was then cutting-edge mobile robotics inside a cramped, retrofitted utility vehicle. Watching it evokes a sense of nostalgic technological ingenuity, capturing the milestone moment when bulky early computers and experimental sensors first took to the open road.

## Bulletin - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) Carnegie Mellon University's 1986 NavLab 1 pioneered onboard autonomous computing by housing researchers and processing hardware entirely within a modified commercial truck. - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) The self-contained mobile architecture bypassed the severe telemetry bottlenecks and communications failures that hindered stationary offboard robotics testbeds. - [01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) NavLab's multimodal perception system integrated color video imagery with scanning laser rangefinder data to construct unified 3D models with geometry and material reflectance. - [01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) A centralized Blackboard software architecture orchestrated the vehicle's asynchronous distributed modules across sensing, mapping, planning, and driving.

## Key takeaways - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) NavLab was developed by CMU's Robotics Institute as an experimental testbed for autonomous driving, image understanding, and human-robot interaction. - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) The platform was built into a roadworthy commercial truck outfitted as a fully self-contained mobile navigation habitat. - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) Housing computing equipment and researchers onboard eliminated external telemetry bottlenecks and dependence on fixed infrastructure. - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) The cabin integrated dedicated operator consoles, internal sensors, and controllers to allow real-time parameter tuning and data analysis. - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) Standard mechanical driving controls were retained so human operators could immediately take over steering and propulsion when necessary. - [01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) Video cameras captured color and intensity imagery to enable algorithmic segmentation of road boundaries and surface textures. - [01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) An active scanning rangefinder used laser distance measurements to capture high-fidelity 3D spatial terrain profiles. - [01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) The laser system recorded surface reflectance alongside geometric depth to enhance environmental characterization. - [01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) NavLab fused color, intensity, range, and reflectance streams into a unified world model for natural scene interpretation. - [01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) A Blackboard software architecture coordinated the distributed subroutines that handled sensing, mapping, path planning, and physical actuation.

## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) Overview and Physical Architecture of the NavLab

Carnegie Mellon University's Robotics Institute developed the NavLab as an experimental testbed dedicated to autonomous navigation, image understanding, and exploring how humans interact with intelligent mobile systems. Built inside a heavily modified, roadworthy commercial truck, the vehicle serves as a completely mobile navigation habitat. By bringing both the research team and substantial computing power directly inside the vehicle, the system circumvents the traditional pitfalls of early autonomous robotics. As the narrator explains, “Because it is self-contained, the NavLab is not subject to telemetry bottlenecks, communication faults, or dependence on stationary infrastructure, and can travel to confront navigation problems at a test site.”

Inside the vehicle shell, the platform integrates all required hardware, including onboard computers, specialized controllers, telemetry gear, and internal sensors. The interior functions as an onboard laboratory where researchers can monitor telemetry, analyze experimental data, and supervise autonomous operations from an operator's console. For safety and operational flexibility, the vehicle retains standard manual driver controls, allowing a human operator to take over steering and propulsion whenever necessary. As the narrator notes, “The Nav Lab is a roadworthy truck modified so that researchers or computers can control the vehicle as occasion demands.”

- **Core Research Domains:** Robot navigation, computer vision/image understanding, and human-machine interaction with intelligent systems. - **Onboard Cabin Capabilities:** Houses internal computing infrastructure, hardware controllers, monitoring consoles, and passenger workspace for data collection and runtime parameter tuning. - **Human Override Functions:** Manual steering and pedal controls allow human intervention for overland travel between sites, initial system setup, and immediate recovery from experimental errors.

---

[01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) Sensor Suite, Perception, and the Blackboard Architecture

To accommodate a wide spectrum of autonomous vehicle research, the NavLab is outfitted with multimodal sensing hardware. Video cameras capture color and intensity imagery to facilitate scene interpretation. Algorithms analyze road boundaries and surface changes by segmenting intensity, texture, and color variations. In parallel, an active scanning rangefinder sweeps the surrounding terrain using a distance-measuring laser, generating high-fidelity three-dimensional spatial data that captures both the geometric profile and the surface reflectivity of the environment.

Rather than relying on isolated streams, the vehicle combines these disparate perceptual feeds into a coherent world model. The narrator emphasizes, “Taken together, color, intensity, range, and reflectance data provide a rich basis for building natural scene descriptions.” To tie these components together computationally, the platform utilizes a centralized Blackboard software framework. The narrator summarizes this control structure by stating, “A Blackboard computer architecture integrates the distributed processes that sense, map, plan, and drive.”

```mermaid flowchart TD accTitle: NavLab Autonomous Architecture accDescr: Data flow from raw sensor inputs through perceptual fusion into the Blackboard architecture to control driving.

subgraph Sensors["Sensing Modalities"] CAM["Video Cameras<br/>(Color & Intensity)"] LIDAR["Scanning Rangefinder<br/>(Laser Range & Reflectance)"] end

subgraph Perception["Data Fusion"] FEAT["Road Segmentation &<br/>3D Geometric Models"] end

subgraph Core["Blackboard Architecture"] BB[("Blackboard Shared Memory")] MAP["Mapping Process"] PLAN["Planning Process"] DRIVE["Driving & Actuation"] end

CAM --> FEAT LIDAR --> FEAT FEAT --> BB BB <--> MAP BB <--> PLAN PLAN --> DRIVE ```

- **Perceptual Processing:** Video inputs identify lane margins and obstacles using color, texture, and intensity segmentation. - **Laser Rangefinding:** Provides geometric depth measurements alongside material reflectance to understand surface profiles in 3D. - **Distributed Software Framework:** The Blackboard system coordinates asynchronous, distributed subroutines responsible for sensing, mapping, path planning, and physical actuation.

## Notable quotes > [01:14](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=74s) Taken together, color, intensity, range, and reflectance data provide a rich basis for building natural scene descriptions. — narrator

> [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) Because it is self-contained, the NavLab is not subject to telemetry bottlenecks, communication faults, or dependence on stationary infrastructure, and can travel to confront navigation problems at a test site. — narrator

## Mentioned Carnegie Mellon University, Robotics Institute, NavLab, NavLab 1, Blackboard

## Video description NavLab or Navigation Laboratory was the first self-driving car with people riding on board. It was very slow, but for 1986 computing power, it was revolutionary. NavLab continued to lay the groundwork for Carnegie Mellon University's expertise in the field of autonomous vehicles.

https://en.wikipedia.org/wiki/Navlab

[Source](https://www.youtube.com/watch?v=ntIczNQKfjQ)

中文简报

卡内基梅隆大学部署NavLab卡车测试车载自动驾驶 直接将计算硬件部署在车内消除了遥测瓶颈并摆脱了对外部基础设施的依赖，使集中式黑板架构能够融合传感器输入并控制行驶。 该视频概述了卡内基梅隆大学NavLab的物理与计算架构，该平台是一辆基于改装商用卡车打造的实验性自动驾驶测试车。通过将计算硬件和研究团队同时搭载在车上，该车辆避免了早期的遥测瓶颈，并允许研究人员在保留人工驾驶接管功能的同时直接前往远程测试地点。NavLab融合了多模态传感器输入——包括彩色与灰度摄像机以及主动扫描式激光测距仪——以生成细致的自然场景描述。这些感知输入、建图系统、路径规划器和行驶程序通过一个集中式黑板计算机架构进行集成，该架构协调着整个自动驾驶管线。 卡内基梅隆大学1986年的NavLab 1开创了车载自动计算的先河，它将研究人员和处理硬件完全部署在一辆改装的商用卡车内部。 (0:00) 这种自给自足的移动架构避开了严重的遥测瓶颈以及通信故障，这些问题曾阻碍了固定的非车载机器人测试台。 (0:00) NavLab的多模态感知系统将彩色视频图像与扫描激光测距仪数据相融合，构建包含几何结构和材质反射率的统一3D模型。 (1:14) 集中式黑板软件架构协调了车辆在感知、建图、规划和行驶等方面的异步分布式模块。 (1:14)

Detailed notes — zh

## Background 该视频由 CMU 机器人研究所（CMU Robotics Institute）于 2016 年 10 月 13 日发布在 YouTube 平台上。视频简介指出，NavLab（导航实验室）是第一辆载人行驶的自动驾驶汽车，尽管以 1986 年的算力水平来看它的行驶速度非常缓慢，但在当时具有革命性意义，并为卡耐基梅隆大学在自动驾驶领域的专业积淀奠定了坚实基础。 ## TL;DR 视频回顾了 1986 年卡耐基梅隆大学研制的 NavLab 1 自动驾驶研究平台。该平台基于适航卡车底盘改装而成，采用将计算设备与研究人员直接容纳于车内的“自包含”架构，彻底摆脱了对外部遥测与固定基站的依赖，并支持随时的人工接管。在技术实现上，NavLab 结合了视频摄像机的颜色/光强数据与激光测距仪的三维几何反射信息，依托经典的黑板（Blackboard）系统架构，实现了感知、建图、规划与底层驱动的高效闭环控制。 ## Atmosphere & style 全片洋溢着浓厚而纯正的 1980 年代硬核科研记录片氛围，旁白沉稳、清晰且严谨，带有极强的学院派学术严谨感。画面中老式庞大的机柜、车载监视器与研究人员在狭小车厢内的协同调试，传递出早期计算机视觉与自动化探索阶段特有的拓荒感与厚重历史质感。 ## Bulletin - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) 卡耐基梅隆大学于1986年研发了NavLab 1，这是首辆支持人员随车乘坐的自动驾驶卡车，奠定了CMU在无人驾驶领域的核心技术基础。 - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) NavLab 1采用完全自包含的车载架构，计算、控制与传感设备全部集成在车身内部，彻底摆脱对外部通信基站与遥测网络的依赖。 - [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) 感知系统融合了彩色视频摄像机与三维激光扫描测距仪，通过综合分析颜色、光强、纹理、空间几何与表面反射率构建多模态场景理解。 - [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) 软件层面引入黑板计算机体系结构（Blackboard Architecture），统筹调度感知、建图、路径规划和驱动控制等分布式进程实现自主闭环控制。 ## Key takeaways - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) NavLab 1是CMU开发的核心实验平台，主要用于开展机器人导航、图像理解与人机交互研究。 - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) 实体车辆由具备合法路权改装的卡车底盘构成，设计为一个能够直接容纳研究人员与计算硬件的“移动导航栖息地”。 - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) 支持计算机自动驾驶与人类驾驶员手动控制的双模灵活切换，保证实验灵活性与安全性。 - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) 全自包含架构使测试摆脱遥测带宽瓶颈、通信中断以及对固定基础设施的依赖，可直接深入实地应对导航难题。 - [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) 保留人类驾驶员的最高控制优先级，用于陆路转场、实验环境设置及突发错误状态下的安全人工接管。 - [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) 车厢内部设立了完整的工作站控制台，研究人员可随车实时监控系统状态、调整算法参数并执行错误恢复。 - [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) 配置视频摄像机采集颜色与光强数据，通过纹理与颜色分割等多维视觉算法精确提取道路边缘。 - [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) 搭载扫描测距仪（激光雷达），扇形扫描获取周边环境的三维几何结构和表面反射率信息。 - [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) 多传感器数据融合显著提升了环境建模与场景理解在复杂自然环境下的鲁棒性。 - [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) 软件系统采用黑板架构（Blackboard Architecture），作为中心枢纽高度协同感知、建图、规划与底盘驱动等多任务进程。 ## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) NavLab 的定位与自包含移动实验平台设计 卡耐基梅隆大学（Carnegie Mellon University）研发了 NavLab（导航实验室），将其作为开展机器人导航、图像理解以及人类与智能系统交互作用研究的核心测试平台。NavLab 的实体是一辆经过改装、具备合法上路行驶能力的卡车，系统设计非常灵活，研究人员与计算机可根据实际需求随时切换对车辆的控制权。作为一个“移动导航栖息地”（mobile navigation habitat），它在车身内部直接容纳了参与测试的研究人员以及庞大的车载计算设备。 这种完全自包含（self-contained）的架构设计带来了巨大的技术优势。以往的无人驾驶或机器人测试往往依赖远程基站通信，而 NavLab 将所有计算与处理能力集中于车内，彻底摆脱了外部限制。旁白特别强调了这种独立性的核心价值：“Because it is self-contained, the NavLab is not subject to telemetry bottlenecks, communication faults, or dependence on stationary infrastructure, and can travel to confront navigation problems at a test site.”（“因为它是完全自包含的，NavLab 不会受到遥测瓶颈、通信故障或对固定基础设施依赖的困扰，并且可以直接前往测试场地去面对并解决现场的导航难题。”）此外，车内保留了完备的驾驶员控制装置，在陆路转场行驶、测试环境设置以及发生实验错误需要恢复时，人类监控人员能够随时覆盖自动控制系统进行人工接管。 - **平台形态**：由适航卡车底盘改装而成的移动导航实验室。 - **控制模式**：支持计算机自主控制与人类驾驶员手动控制双模切换。 - **系统自包含性**：计算与传感全集成于车体内，无需外部固定基站支持，免疫通信中断与遥测带宽受限问题。 - **安全保障机制**：人工驾驶装置具备最高控制优先级，可用于转场移动、系统配置及实验异常接管。 ### [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) 车载工作空间、多模态传感与黑板架构系统集成 NavLab 的车厢外壳不仅妥善安置了包括车载计算机、控制器、遥测装置和内部传感器在内的所有核心硬件，还为测试人员提供了一个功能完整的随车工作区。研究人员可以直接坐在车辆内部收集、分析一手数据，并通过操作员控制台对整套系统进行全程监控与督导，实时完成参数微调、实验配置与错误恢复。为了满足多样化导航研究的需求，NavLab 配备了极其丰富的感知硬件选择。其中，视频摄像机负责采集彩色与光强图像以进行场景理解，例如通过结合光照强度、纹理特征与颜色分割算法来精确识别道路边缘；而安装的扫描测距仪（激光测距雷达）则对车身周围环境进行扇形扫描，提供有关周围环境三维几何形状与表面反射率的关键空间信息。 在软件与系统架构层面，系统将来自多个传感源的图像、距离与反射数据融合在一起，从而大幅提升环境感知的鲁棒性。正如旁白所指出的：“Taken together, color, intensity, range, and reflectance data provide a rich basis for building natural scene descriptions.”（“综合来看，颜色、光强、测距以及反射率数据为构建自然场景描述提供了坚实丰富的基础。”）为了有效协调这些纷繁复杂的软硬件模块，NavLab 采用了黑板（Blackboard）计算机体系结构，将感知（sense）、建图（map）、路径规划（plan）与车辆驾驶（drive）等一系列分布式计算进程高度整合，实现了自主闭环控制。 ```mermaid flowchart TD accTitle: NavLab 1 系统感知与控制架构 accDescr: 展示 NavLab 1 从多传感器采集、数据融合到黑板架构规划驱动的完整数据流 subgraph Sensors ["感知采集层"] CAM["视频摄像机<br/>(颜色 / 强度图像)"] LIDAR["扫描测距仪<br/>(3D几何 / 反射率)"] end subgraph Processing ["数据融合与场景理解"] FUSION["多源传感器融合<br/>(道路边缘检测/环境建模)"] end subgraph Architecture ["黑板体系结构 (Blackboard)"] BB["黑板中央信息共享"] P_SENSE["感知进程 (Sense)"] P_MAP["建图进程 (Map)"] P_PLAN["规划进程 (Plan)"] P_DRIVE["驾驶进程 (Drive)"] end subgraph Control ["执行与监控层"] VEHICLE["车辆底盘执行机构"] CONSOLE["操作员控制台 / 人工接管"] end CAM --> FUSION LIDAR --> FUSION FUSION --> BB BB <--> P_SENSE BB <--> P_MAP BB <--> P_PLAN BB <--> P_DRIVE P_DRIVE --> VEHICLE CONSOLE -.->|"实验调试 / 超控干预"| VEHICLE ``` - **传感硬件组合**：彩色视频摄像机（提取颜色与强度信息）搭配激光扫描测距仪（获取三维距离与反射率）。 - **感知算法逻辑**：道路边缘提取依赖光强、纹理与颜色分割的多特征综合分析；三维激光数据补充空间几何轮廓。 - **系统集成枢纽**：采用黑板架构（Blackboard Architecture）整合分布式任务流，协同调度感知、建图、规划与底层驱动。 - **交互与运维**：研究员可在舱内控制台完成参数调优（tuning）、系统设置（setup）与异常排查（error recovery）。 ## Notable quotes > [00:00](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=0s) Because it is self-contained, the NavLab is not subject to telemetry bottlenecks, communication faults, or dependence on stationary infrastructure, and can travel to confront navigation problems at a test site. > [00:49](https://www.youtube.com/watch?v=ntIczNQKfjQ&t=49s) Taken together, color, intensity, range, and reflectance data provide a rich basis for building natural scene descriptions. ## Mentioned NavLab, NavLab 1, Carnegie Mellon University, CMU Robotics Institute ## Video description NavLab or Navigation Laboratory was the first self-driving car with people riding on board. It was very slow, but for 1986 computing power, it was revolutionary. NavLab continued to lay the groundwork for Carnegie Mellon University's expertise in the field of autonomous vehicles. https://en.wikipedia.org/wiki/Navlab [Source](https://www.youtube.com/watch?v=ntIczNQKfjQ)

Source quotes

[0:00] Because it is self-contained, the NavLab is not subject to telemetry bottlenecks, communication faults, or dependence on stationary infrastructure, and can travel to confront navigation problems at a test site. [0:49] Taken together, color, intensity, range, and reflectance data provide a rich basis for building natural scene descriptions.

Source quotes

[1:14] narrator: Taken together, color, intensity, range, and reflectance data provide a rich basis for building natural scene descriptions. [0:00] narrator: Because it is self-contained, the NavLab is not subject to telemetry bottlenecks, communication faults, or dependence on stationary infrastructure, and can travel to confront navigation problems at a test site.

Original transcript

[0:00] We have recently built the Nav Lab, a [0:02] test bed for research in robot [0:04] navigation, image understanding, and the [0:06] role of human interaction with [0:09] intelligent systems. The Nav Lab is a [0:12] roadworthy truck modified so that [0:14] researchers or computers can control the [0:17] vehicle as occasion demands. As a mobile [0:20] navigation habitat, it accommodates [0:22] researchers and significant computing on [0:24] board. Because it is self-contained, the [0:27] navl lab is not subject to telemetry [0:30] bottlenecks, communication faults, or [0:32] dependence on stationary infrastructure, [0:35] and can travel to confront navigation [0:37] problems at a test site. Driver controls [0:40] allow a human monitor to override [0:42] automatic control for overland travel, [0:45] setup, and recovery from experimental [0:48] errors. [0:49] The NavLab shell houses all onboard [0:52] equipment, including computers, [0:54] controllers, telemetry, and internal [0:57] sensors. In addition, it provides a [0:59] working area for operators and allows [1:02] researchers to gather data within the [1:03] confines of the [1:05] vehicle. Researchers can monitor and [1:08] supervise the Nav Lab from the [1:10] operator's console for setup, error [1:12] recovery, and tuning. The Nav Lab [1:15] supports a choice of sensing to [1:16] accommodate many types of navigation [1:18] research. Video cameras provide color [1:21] and intensity images for scene [1:23] interpretation. Road edges, for example, [1:26] are analyzed through intensity, texture, [1:28] and color [1:29] segmentation. A scanning rangefinder [1:32] sweeps the surroundings with a distance [1:34] measuring laser that provides useful [1:36] three-dimensional information about the [1:39] geometry and reflectivity of the [1:41] environment. [1:42] Taken together, color, intensity, range, [1:45] and reflectance data provide a rich [1:47] basis for building natural scene [1:49] descriptions. Sensor information from [1:51] several sources can be fused to achieve [1:54] more robust perception. A Blackboard [1:57] computer architecture integrates the [1:58] distributed processes that sense, map, [2:02] plan, and drive.