亡灵机器人学：死去的身体，活着的机器

2022年，莱斯大学的研究人员把一根针插入死狼蛛的头胸部，用胶固定，接上注射器。加压时，八条腿张开；释放压力，腿便合拢，夹起物体。演示中，它从面包板上拔出跳线，使LED熄灭，还抓起了另一只蜘蛛。莱斯大学将这项研究称为“necrobotics”，即利用死去生物的结构制作机器人部件。

第一作者Faye Yap当时是研究生，在Daniel Preston的实验室工作。论文于2022年7月发表于《Advanced Science》，研究团队次年获得搞笑诺贝尔奖。这不是让蜘蛛复活：针头和外部气源控制腿的开合，蜘蛛的神经系统并未恢复工作。

蜘蛛腿有使其向内收拢的屈肌，伸展则依赖体内的液压。蜘蛛死后失去主动加压能力，腿往往蜷曲。莱斯团队以外部气压驱动保留下来的结构，做成夹爪。论文测试显示，它能提起约为自身质量1.3倍的物体，但约700次开合后已开始退化；1000次测试结束时，部分关节膜出现裂纹。研究人员认为，脱水可能使关节变脆。

蚊子口器和甲壳类外骨骼

麦吉尔大学与德雷塞尔大学随后试验了另一种材料：雌蚊的口器。研究人员在显微镜下取出口器，把它固定在塑料点胶头上，再装入3D打印装置。这项“necroprinting”研究于2025年11月在线发表于《Science Advances》。

论文报告的最细打印线宽为20微米，比较对象是商用36号点胶头，并非所有3D打印技术。麦吉尔大学的说明列出蜂窝、枫叶及包裹细胞的支架等打印演示，也提到清洁后重复使用的可能性。这些成果属于制造实验，不能据此称它已经能用于人体治疗或临床手术。

瑞士洛桑联邦理工学院（EPFL）的研究使用挪威海螯虾（langoustine）腹部外骨骼。它由硬壳和柔性关节膜组成，研究人员加入人造弹性材料、牵引机构和涂层，再连接电机，制作抓取及游泳装置。论文首次在线发表于2025年11月，编入2026年期刊卷次。

这里的“3克承载680克”是外骨骼在特定高刚度姿态下、达到失效前的承载测试，不是完整机器人能举起自身质量226倍的物体。EPFL报告，装置能够移动500克物体，游泳机器人速度最高为每秒11厘米。外骨骼来自食品加工材料，但电机、控制系统及人造材料仍是装置的一部分，不能把整台机器人称为可生物降解。

不同实验的指标与条件

来源：蜘蛛论文、口器喷嘴论文摘要、外骨骼论文。各项指标采用论文测试口径，不作跨装置性能排名。

蛾触角无人机采用的是另一种工作方式。信州大学和千叶大学2025年公布的装置依靠保有活性的家蚕蛾触角接收气味信号；改进传感器和搜索算法后，在实验中抵达5米外的气味源，传感器寿命延长至约5小时。两校公告称其为生物混合无人机。它需要触角组织保持活性，与死体结构的机械利用不同，搜救等用途仍是研究方向。

使用寿命、成本与取材

这些装置已经完成抓取、打印和运动演示，长时间使用仍有具体困难。蜘蛛关节会脱水；海螯虾外骨骼的个体形状不同，配成夹爪后，两侧弯曲也可能不一致。EPFL研究人员提出，需要进一步改进控制机构。天然材料省去部分加工步骤，并不等于成品免费：蚊子口器需要显微取材和安装，外骨骼需要加装驱动部件，使用寿命也会影响更换频率。

如果用于组织工程制造，还须分别验证打印材料、细胞存活和设备稳定性。打印出包裹细胞的支架，与证明该支架能够安全用于患者，是不同阶段。现有实验数据也不能直接推算销售收入，更不能把整个生物混合机器人或软体机器人行业的预测规模算到necrobotics名下。

材料来源同样有区别。莱斯团队论文记载，实验狼蛛经低温处理致死；麦吉尔大学说明，蚊子来自合作机构用于生物研究的实验室种群，经安乐死后取材；EPFL则使用已经进入食品加工流程的海螯虾外骨骼。使用既有加工废料与为实验专门获取动物，不能采用同一套伦理说明。

英国《动物福利（感知能力）法》在2022年已将十足目甲壳类纳入其动物定义，用于政府政策的福利考量。2026年9月10日，英国内政部在议会书面答复中表示，科研使用方面将分阶段支持基本福利标准及自愿报告，补充证据和技术研究，再为未来是否将十足目甲壳类纳入《1986年动物（科学程序）法》的受保护动物提供依据。答复并未宣布后者已实施。

在EPFL的装置中，使用后的外骨骼可以与机器人底座分离，大部分人造部件可再利用。这是研究团队提出的回收方式；天然外骨骼的个体差异及使用后的材料处理，仍须纳入装置设计。

修订说明：2026年10月6日，本文核对原始研究后修正研究者身份、蜘蛛驱动机制及测试指标，删除无法作为necrobotics规模依据的市场预测和商业化时间表，并补入英国2026年9月的官方答复。首发日期不变。

English version

In 2022, researchers at Rice University inserted a needle into the body of a dead wolf spider, sealed it with glue and connected it to a syringe. Pressure opened the eight legs; releasing it closed them around an object. Demonstrations included removing a jumper wire from a breadboard to switch off an LED and lifting another spider. Rice called the approach “necrobotics”: using structures from dead organisms as robotic components.

Lead author Faye Yap was a graduate student in Daniel Preston’s laboratory. The study appeared in Advanced Science in July 2022, and the team received an Ig Nobel Prize the following year. The apparatus did not revive the spider or restore its nervous system. An external air supply moved the legs.

Spiders have flexor muscles that curl their legs inward; hydraulic pressure extends them. Death removes their ability to pressurize the body actively. Rice’s tests showed that the resulting gripper could lift about 1.3 times its own mass. Degradation began after roughly 700 cycles, rather than only after 1,000. Cracks appeared in some joint membranes by the end of the 1,000-cycle test, possibly because dehydration made them brittle.

Mosquito nozzles and crustacean shells

A McGill–Drexel team used female mosquito proboscides as printing nozzles. Researchers removed the feeding tubes under a microscope, attached them to plastic dispensing tips and installed them in a 3D-printing system. Their “necroprinting” paper was published online in Science Advances in November 2025.

The finest reported line width was 20 micrometres. The comparison was with commercial 36-gauge dispensing tips, not every form of 3D printing. McGill described demonstrations including honeycomb structures and cell-containing scaffolds, and said the nozzles could be reused with appropriate handling and cleaning. These were manufacturing experiments, not evidence of a clinical treatment or surgical application.

EPFL researchers used abdominal exoskeletons from langoustines, also known as Norway lobsters. They combined the jointed shells with synthetic materials, tendons, coatings and motorized mechanisms to build gripping and swimming devices. The paper first appeared online in November 2025 and was assigned to a 2026 journal volume.

The often-cited 680-gram load was a failure-limit measurement for a three-gram exoskeleton in a particular high-stiffness configuration. It was not a complete robot lifting 226 times its own mass. EPFL reported a manipulator moving a 500-gram object and a swimming robot reaching 11 centimetres per second. The shells came from food-processing material, but the motors and synthetic components were not thereby biodegradable.

Experimental results and measurement limits

Sources: spider study, nozzle abstract, exoskeleton study. Results use each study’s own test conditions and are not a performance league table.

Moth-antenna drones work differently. Shinshu and Chiba universities’ 2025 system used biologically active silkworm moth antennae to detect odours. Revised sensors and a search algorithm helped it reach an odour source five metres away in experiments; sensor lifetime reached about five hours. The universities’ announcement described a biohybrid drone. Its sensing tissue must remain active, unlike the dead mechanical structures; rescue applications remain a research objective.

Durability, cost and sourcing

The experiments demonstrate gripping, printing and movement, but practical limitations remain. Spider joints dry out. Langoustine shells vary naturally, so the two fingers of a gripper may bend differently; EPFL researchers said better control mechanisms would be needed. Avoiding some fabrication steps does not make a finished device free. Mosquito nozzles require delicate preparation and installation, shells need drive mechanisms, and replacement frequency depends on durability.

Tissue-engineering applications would also require separate evaluation of printing materials, cell survival and equipment reliability. Producing a cell-containing scaffold does not establish that it is safe for a patient. Nor do laboratory performance figures establish sales revenue or justify treating forecasts for all biohybrid or soft robotics as a necrobotics market estimate.

Sourcing matters. Rice’s paper describes killing the spiders through cold exposure. McGill said its mosquitoes came from a partner institution’s biological-research colonies and were euthanized before collection. EPFL used shells from animals already processed for food. Recovering existing waste and obtaining animals specifically for an experiment require different ethical explanations.

The UK’s Animal Welfare (Sentience) Act 2022 includes decapod crustaceans within its definition of animals for consideration of welfare in government policymaking. In a written parliamentary answer on 10 September 2026, the Home Office described a phased approach involving baseline welfare standards, voluntary reporting and work on evidence and technical gaps. This would inform any future consideration of including decapods as protected animals under the Animals (Scientific Procedures) Act 1986. The answer did not announce that such an extension had been implemented.

EPFL’s proposed recycling process separates the used shells from the robotic base so that most synthetic components can be reused. Natural variation and disposal of the biological material remain design considerations.

Correction and update, 6 October 2026: This article corrects the researcher’s status, spider actuation mechanism and experimental measurements, removes unsupported necrobotics market estimates and commercialization timelines, and adds the UK’s September 2026 official response. The original publication date is unchanged.