Top 9 Innovations in Robotics and AI at FIX 2026: Discover the Future

I flew to Daegu for FIX 2026. The venue hummed with engineers and investors. I saw nine companies collect a top innovation prize. The panel chose them from 209 applicants. Twenty-six judges weighed the field. Another 22 companies received Innovation Awards. That brought the total winners to 31. Six winners came from Daegu-Gyeongbuk. You get a clear view of Korea’s tech pipeline. I interviewed two founders between demos. South Korean builders drove many demos.

Robotics took center stage in the hall. Tesollo showed its DG-5F-M robotic hand. The device has five fingers. It offers 20 degrees of freedom. Degrees of freedom describe movable joints. Each finger joint adds motion. Tesollo says that helps mimic human grip. It also showed a smaller DG-5F-S model. The compact hand weighs 880 grams. The company says it keeps strong manipulation. Humanoid hands remain a hard engineering problem. Fingers need precision, force, and control. Batteries and heat limit every move. Tesollo claims its actuators solve some constraints. Its older three-finger hand won a CES award. That award helps it court industrial buyers.

AeiROBOT presented Alice 5 on the main stage. I watched it walk from the demo area. The robot stands about 170 cm tall. It weighs around 50 kg. It has 43 degrees of freedom. AeiROBOT says that gives fluid motion. Alice 5 uses NVIDIA’s Jetson Thor platform. The company mixes actuator types across the body. Upper limbs use quasi-direct-drive motors. This helps the arms move quickly. Legs use linear actuators for stability. This design tries to balance strength and responsiveness. Wheeled robots move well on flat floors. Legged robots can handle stairs and clutter. Alice 5 also supports interchangeable hands. That lets one robot change tasks. AeiROBOT is testing the line in factories. Earlier Alice M1 units worked in PCB plants. Alice 5 targets shipyards operated by HD Hyundai. Hanwha Ocean is also involved in tests. The robots may fight fires or clean hazardous areas. The company aims for broad industrial use by 2028.

Demos often look cleaner than real factories. See IEEE Spectrum at https://spectrum.ieee.org for robot news. Wired has covered robot hand advances. Read Wired at https://www.wired.com for hand news. These reports give useful context. You can search robotika innovációk FIX 2026. They do not replace independent tests.

Mobility startups took the next slot. AI-native tools link data to action. Onepredict showed its AI Native Factory package. Traditional maintenance systems flag possible machine failures. Onepredict connects equipment health to production plans. It helps managers decide when to repair. It helps them choose what to fix first. The value comes from joining two data streams. A broken machine can ripple through a line. Early warnings matter more than dashboards. MIT Technology Review explains this factory shift. Read MIT Technology Review at https://www.technologyreview.com for factory AI. Its articles link AI to shop-floor decisions. It should not stop at alerts.

SWM pitched an urban robotaxi model. A robotaxi is a self-driving taxi. The company runs driver-assist services in Gangnam. It already has paid commercial rides. A safety driver still sits onboard. Its AP-700 platform delivers 2,000 TOPS. TOPS measures AI operations per second. That speed helps sensor analysis. SWM says flexible routes attract city riders. Fixed-route shuttles often feel too rigid. Real roads bring unpredictable pedestrians and cyclists. Level 4 means the car handles most driving tasks. It still needs careful validation. I asked a panelist about insurance. Regulators watch route changes according to a panelist. SWM needs more public data to build trust.

Perseus won for PEGASUS, a hypervisor for cars. Modern cars run many software domains. A hypervisor splits one chip among software areas. Cockpit systems need rich graphics. Control systems need strict safety timing. Perseus lets them share one chip. PEGASUS keeps separate environments isolated. It earned ISO 26262 ASIL-D certification. That label signals high safety rigor. Automakers want fewer electronic control units. Each extra box adds wiring and cost. One chip can simplify updates. Yet software bugs can cross a wall. Perseus must prove isolation under pressure. Buyers will ask for failure logs.

HJWave brought radar antennas to the stage. The firm makes 3D waveguide antennas. They guide radio waves with less loss. The company says resolution improves. Detection coverage also expands. PCB antennas can lose signal at edges. Waveguide designs may hold power better. Radar helps cars see through rain or dust. Cameras can fail in low light. HJWave holds Korean and US patents. It received a European patent decision. It signed a supply deal with an autonomous firm. The customer stayed unnamed. Better radar helps but does not solve self-driving alone.

Cooling and security filled the late sessions. CENS showed CENS-D130 precision spray cooling. AI servers produce extreme heat. Full immersion can flood a rack with liquid. CENS sprays nonconductive fluid on hot parts. The fluid does not carry electricity. The flow changes with thermal demand. The company says it uses less cooling fluid. That can cut weight and maintenance burden. Data centers need dense power and airflow. Cooling failures still threaten chips and uptime. CENS needs long-term field records. Buyers care about pump noise and leak risk.

WACE presented VEXPLOR, a factory AI operating system. Many plant tools show charts and alerts. VEXPLOR connects decisions to machine controls. It can send instructions to PLCs. PLCs are industrial controllers. Those controllers drive equipment. The company adds a verified safety layer. It hands control to AI one level at a time. Unsafe commands get blocked. This design aims to close the gap. A dashboard tells a human what to do. VEXPLOR tries to let the machine respond. That raises trust questions. Factories move slowly for good reason. Engineers need clear rollback options.

PotatoNet rounded out the winners. CEO KyoungA Shin spoke about hidden web threats. The firm sells OLPEMI, an AI security platform. Phishing pages trick people with fake links. It scans malicious URLs and phishing pages. It can probe deep web infrastructure. Conventional blocklists lag behind new attacks. OLPEMI tries to find risks before clicks. It also watches mobile links and web servers. Proactive threat intelligence can shorten response time. Yet false positives can block useful sites. Buyers will demand clean alerts. The product faces a crowded market. Potatonet says its focus is discovery.

I left Daegu with a practical list:

• Humanoid robots need reliable hands
• Robotaxi safety needs transparent records
• Software-defined cars need certified isolation
• Radar must cut through fog and rain
• AI factories need controlled autonomy
• Data centers need targeted cooling
• Cyber defenses need early visibility

These nine companies attack real bottlenecks. FIX 2026 showed the work is still unfinished. You should watch their field pilots. Proof may earn the next prize. Promises alone rarely win trust.