Table of contents
Ayming Institute :the think tank of the Ayming Group.
The Ayming Institute (AI) aims to help leaders in the private and public sector gain a deeper understanding of the evolving global economy by focusing on three areas.
The first area is sustainability. We believe that the environment and social responsibility are critical issues for businesses today. For this reason, our content aims to help companies integrate these issues into the way they make decisions.
The second area is business development. Through our content, we wish to help companies to develop a stronger business culture and a sustainable approach to growth.
The third area is the people side of the business. With our content, we want to support individuals as they navigate their careers, learn new skills, and find ways to contribute in a world that is constantly changing.
Our strongest commitment is to help organizations better understand how markets are changing, and how they can build better businesses as a result. We aim to do this by providing analysis of the global economy’s transformation; sharing our insights through thought-provoking publications, and engaging business leaders in conversations about the economic changes that are affecting all of us.
AI & Robotics
Technological development is often a step ahead of general understanding of its implications. With artificial intelligence (AI) the pace and significance are such that policymakers are trailing behind and even tech luminaries are at loggerheds. While one predicts that humanoid robot helpers will outnumber people, another warns that god-like artificial general intelligence (AGI) could destroy the human race.1
However plausible or dystopian the fears that intelligent robots could take over the world, they are certainly more capable of taking on a variety of new tasks. The progress in generative AI, machine learning, deep learning – coupled with advances in computer vision, sensors, positioning and mobility technologies – open up new possibilities. Imbued with creativity and flexibility, AI-powered robotics can make further inroads in and beyond the factory with applications in fields as diverse as agriculture and autonomous vehicles.
Productivity and precision are powerful drivers of AI automation. But there are also the increasingly strong push factors of skills shortages and increased operational costs. With GenAI and other developments, intelligent robotics can complement human ingenuity as well as replace certain types of labour while offering companies a potentially future-proof competitive advantage.
Today, around three million industrial robots are at work worldwide, with around 400,000 new entrants annually. Many are now infused with AI capabilities – whether through machine learning or deep learning algorithms – and a variety of sensors (for vision, vibration, proximity, movement and other environmental aspects) feeding data they can analyze and act upon in real time.2 This global AI robotics market, worth $14.3 billion in 2023, is projected to reach a staggering $82.5 billion by 2032, growing an average 21.5% per year.
From farm to pharma
Intelligent robots are tailor-made for the smart factories of Industry 4.0, but consider the other problems they can solve, from food supply to healthcare.
By 2050, global food demand is projected to rise 35-56%.3 Yet the agricultural industry faces severe field labour shortages, making automation attractive.4 With all-weather functionality, sub-millimetre precision, and 24-hour operation, agricultural AI-powered robotics can automate picking and combat herbicide resistance and fertilizer over-use, not to mention applications from weather and harvest predictions.
The award-winning LaserWeeder from Carbon Robotics employs deep-learning computer vision, Nvidia graphics processing units (GPUs), and 30 industrial CO2 lasers to identify and kill up to 99% of weeds. Laser-weeding can preserve soil microbiology, while cutting weed control costs by up to 80%.5
Robotic harvesting has seen a wave of AI innovation since the 2012 Agrobot E-Series strawberry picker. From Tevel AI orchard drones to Bowery’s Traptic 3D vision robotics,6 intelligent 24/7 harvesting has been implemented both in large fields and vertical farms across the US, Israel, and Europe.7
Most picking robots employ integrated colour sensors and GPUs to assess fruit ripeness in real time, and feed coordinates to a robotic picking arm. However, damage rates are still significantly greater than for human picking. Nevertheless, a handful of companies, such as Harvest CROO Robotics, are increasing the yield of marketable fruits using LiDAR detection and ranging systems (which also enhance collision avoidance in autonomous cars).
In pharmaceuticals, meanwhile, AI should streamline drug development from lab bench to bedside. The healthcare robotics market, worth $4 billion in 2022, is expected to double by 2032.8 Innovation is delivering productivity and precision through automated laboratories9 and cobots like the Mako robotic arm for joint replacement surgery.10 Intelligent ultraviolet disinfection assistants like SAM from Loop Robots can navigate autonomously to sterilize hospital rooms and prevent hospital-acquired infections.
Ever-smarter factories
Back in the factory, AI automation is reinventing the lean manufacturing of the 1990s amid challenging manual labour shortages.11 Empowered by the Internet of Things (IoT), Big Data, machine learning, and other convergent Industry 4.0 technologies, smaller AI cobots are expected to be working alongside more than half of production operators within a decade.12 And that includes the mass of small enterprises, where jobs should be created as well as gaps filled.13
Industrial machining of component parts is typically small-scale with 95% of factories employing fewer than five employees.14 Yet the “world’s largest cottages industry” churns out a trillion dollars’ worth of components each year. Most use outdated computer numerical control (CNC) software requiring trained professionals to enter thousands of operational instructions for each part. However, innovative software by CloudNC can automate up to 80% of instruction input with AI, reducing the burden on staff.
Industrial robots have proved adept at repetitive, programmable tasks. AI (drawing on deep learning and computer vision) can equip the next wave for generic jobs and less structured environments. AI capabilities are already helping to automate ‘piece picking’ in production and distribution facilities, industrial assembly, and manual tasks in industries that had been slow to adapt, such as apparel manufacturing.15
Coupling AI with high-resolution sensors and computer vision could also streamline quality control. Automated assembly in-line monitoring systems, like those provided by Novacura Flow and Intel, can help minimize waste and operational downtime. Audi’s Neckarsulm factory in Germany, for example, has 2,500 autonomous robots but industry- standard end-of-line welding quality control is manual – inspection of one car from the thousand completed each day. In a proof-of-concept collaboration with Intel, machine- learning algorithms and predictive analytics were used to build a scalable model for in-line inspections of welds that can also be applied to other function such as riveting, gluing, and painting.16
Market projections for generative AI in the automotive industry – in vehicle design, production, and autonomous systems – indicate an annual growth rate of 24% from 2023 to 2032.17
A higher level of autonomy
If the technology is taking manufacturing quality standards to a new level, the autonomous vehicle (AV) industry is hoping for a similar boost on its uphill drive to Level 5 where human oversight becomes obsolete.
Despite strides in recent years, most current systems are Level 2 or below (driver assistance, and steering and acceleration control, with human oversight).18 Level 3 ‘conditional automation’ and Level 4 ‘operation without human input in specific environments’ have had limited success.
In October 2023, a Level 4 Cruise robo-taxi in San Francisco failed to anticipate that a parallel incident would throw a pedestrian in its line of travel, lost track of the person and ultimately dragged her beneath the car.19
GenAI could teach AV systems to understand the real world better, plan and anticipate driving actions, and their impacts. Said to be the first research model developed specifically for autonomous driving, GAIA-1 was trained by Wayve on 4,700 hours of UK driving data.20 The aim is to fast-track the development of a Level 4 system.
The industry hopes GenAI can be a tool to create maps in real time without relying on pre- mapped data, and make autonomous systems more robust.21 Beyond that, it could help Level 5 vehicles break free from geofencing restrictions, go anywhere and perform any driving task, just like experienced human drivers. Generative models would be used for decision-making, route planning, understanding complex urban environments, and even simulating rare and extreme situations.22
Germany’s 2021 autonomous driving act permits Level 4 operation of AVs within designated areas with technical supervision.23 A significant milestone in Europe’s legal framework for autonomous driving, it set the stage for future international regulation while also bolstering the country’s lead in technological development.
Public transport and trucking are expected to be the first beneficiaries. The goal of the ATLAS-L4 R&D project, funded by the German Federal Ministry of Economics and Climate Protection, is working towards the first test of a driverless truck in live traffic on a German highway.24 By the end of 2024, the MAN truck (with a safety driver) will travel short sections of the A9 under remote supervision from a control centre, with the eventual goal of covering the entire 166km Munich-Nuremberg route.25
Ongoing progress and innovations will rely on advances in 5G technology to provide reliable digital infrastructure for the success of connected automated driving.26
Towards total connectivity
With 5G coverage offering speeds up to 10 times faster than its predecessor,27 the potential of connected and automated driving may be fully realized.28 Telekom’s initial 5G tests on the A9 freeway between Nuremberg and Munich showed impressive results, achieving latencies of less than 20 milliseconds. Low latency is essential for ensuring seamless communication for AVs to process and receive data and traffic updates in real time, and facilitate remote monitoring and control if required.
Integration of networking technologies will play a crucial role in enhancing road safety and efficiency. The advent of 5G (and ultimately 6G) connectivity enables AV development through reliable vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, and eventually vehicle-to-everything (V2X). Intellias, a V2X developer, is striving to expand communication beyond vehicles to include pedestrians, cyclists, and other parts of the mobility ecosystem.
Greater network connectivity is the driver for advanced warning systems, reliable monitoring, and improved safety provision in AVs.29 However, with connectivity comes risk. As future vehicles become more vulnerable to hacker attacks, safety as well as data privacy could be compromised.30
Manufacturers have to take account of these risks at an early stage of development. As well as rigorous security testing, common technical standards are essential to foster consumer confidence and technological innovation. The German auto trade organization VDA hopes this can be achieved through the Catena-X automotive network, which is developing a standard for electronic data exchange.31 The aim is to create a German basis for a European solution rather than a mandatory legal framework.
While the timescale for autonomous operation has slipped worldwide, North America and China are also vying for first-mover advantage. Fully autonomous trucking is expected to reach viability between 2028 and 2031, with commercial L4 robo-taxis expected to operate at scale by 2030.32
Limitations and lags
However bullish the predictions for AI-empowered robots and AVs, we need to bear in mind the technical, regulatory and ethical limitations.
AI always raises a trust issue; even the algorithm architects don’t really know how decisions made in the ‘black box’ are reached. GenAI’s liability to hallucinate compounds the problem. Also, over time, models degrade if they cannot adjust to new data and need to be retrained. While 85% accuracy may be sufficient for chatbots and other tasks, it is unacceptable when automating decisions on personal loan applications or operating expensive machinery in a factory (or vehicles on the highway).
This opens a huge gap in business use cases that an alternative AI could fill. Start-ups are taking two main approaches to the problem: building a different architecture or teaching pre-trained models to self-correct.33 For example, in the UK, Umnai is combining the neural nets that power large models with rule-based logic, while Conjecture breaks down AI systems into separate processes that businesses could combine for specific tasks, while ensuring the system was reliable, auditable and fixable. Aligned AI, meanwhile, is developing systems to improve existing models through human feedback and teaching them to generalise and extrapolate from rules.
While AI develops and is deployed at breakneck speed, vehicle automation faces different impediments. As well as concerns over public safety and acceptance, the pace of statutory regulation and standards development act as a brake on progress.
1 https://fortune.com/2023/03/02/elon-musk-tesla-a-i-humanoid-robots-outnumber-people-economy/
2 https://www.intel.com/content/www/us/en/robotics/artificial-intelligence-robotics.html
3 https://www.nature.com/articles/s43016-021-00322-9
4 https://www.europarl.europa.eu/RegData/etudes/STUD/2019/629209/IPOL_STU(2019)629209_EN.pdf
5 https://www.businesswire.com/news/home/20230621076749/en/Carbon-Robotics
6 https://techcrunch.com/2022/02/16/following-acquisition-by-bowery-traptics-strawberry-picking-robotics-move-into-vertical-farming/
7 https://www.wired.com/story/elusive-hunt-robot-pick-ripe-strawberry/
8 https://www.precedenceresearch.com/healthcare-service-robots-market
9 https://new.abb.com/news/detail/110608/prsrl-abb-robotics-partners-with-xtalpi-to-build-intelligent-automated-laboratories
10 https://www.stryker.com/us/en/portfolios/orthopaedics/joint-replacement/mako-robotic-arm-assisted-surgery.html
11 https://ifr.org/post/up-skilling-today-and-tomorrows-workforce-for-automation
12 https://hbr.org/2023/11/a-new-generation-of-robots-can-help-small-manufacturers
13 https://ifr.org/ifr-press-releases/news/next-generation-skills
14 https://sifted.eu/articles/ai-manufacturing
15 https://www.controleng.com/articles/industrial-robots-powered-by-ai-improve-manufacturing/
16 https://www.intel.com/content/www/us/en/customer-spotlight/stories/audi-automated-factory.html
17 https://www.precedenceresearch.com/generative-ai-in-automotive-market
18 https://www.digitimes.com/news/a20231121PR201/autonomous-driving-av-tesla-waymo.html
19 https://www.sfchronicle.com/projects/2024/cruise-sf-collision-timeline/
20 https://www.theengineer.co.uk/content/news/wayve-announces-first-generative-ai-model-for-autonomous-driving/
21 https://www.leewayhertz.com/generative-ai-in-automotive-industry/#Generative-AI’s-impact-across-different-levels-of-vehicle-automation
22 https://www.leewayhertz.com/generative-ai-models/
23 https://www.ippi.org.il/germany-autonomous-driving-act/
24 https://d1619fmrcx9c43.cloudfront.net/fileadmin/corporate/press/releases/2023/leoni-press-release-atlas-l4-project-20231024.fin.pdf?1698161962
25 https://vision-mobility.de/news/man-atlas-l4-autonom-auf-der-autobahn-286044.html
26 https://www.vda.de/en/press/press-releases/2022/220520_PM_Germany-secures-pole-position-in-autonomous-driving
27 https://technologymagazine.com/articles/how-6g-networks-will-transform-enterprises
28 https://www.vda.de/en/topics/digitization/5g-ausbau
29 https://www.assemblymag.com/articles/98013-v2v-and-v2x-technology-paves-the-way-for-autonomous-driving.
30 https://www.tuvsud.com/en/-/media/global/pdf-files/infographics/stories/tuvsud-autonomous-connected-cars-hr.pdf
31 https://www.vda.de/en/topics/digitization/automatisiertes-fahren
32 https://www.mckinsey.com/features/mckinsey-center-for-future-mobility/our-insights/autonomous-vehicles-moving-forward-perspectives-from-industry-leaders
33 https://sifted.eu/articles/ai-has-a-trust-problem-meet-the-startups-trying-to-fix-it
