![](\"\/wp-content\/uploads\/2025\/09\/RPF-lumerobotics-2025-03-info1-ING-DESK.png\")
\n <\/picture>\u201cCompanies want to optimize operations and reduce costs,\u201d says computer engineer R\u00e2nik Guidolini, executive director of Lume. He points out that autonomous navigation can offer savings of up to 17% in fuel consumption and 75% in labor, in addition to reducing accident rates by up to 94%. \u201cEven the most experienced driver is not always able to change gears at the right time to drive in the most economical and efficient way,\u201d he explains.<\/p>\n
The company’s current direction results from the success story of the Intelligent Autonomous Robotic Car (IARA), which was developed by a team that included Guidolini at UFES\u2019s Laboratory for High Performance Computing (LCAD). In 2017, the vehicle traveled 74 kilometers (km) between the university campus and the city of Guarapari, in the Greater Vit\u00f3ria Metropolitan Region, crossing three municipalities on the way and navigating daily traffic, toll booths, and traffic lights. A driver was seated in the autonomous car during the journey, ready to take control in the event of any problems, but none occurred.<\/p>\n
In 2018, while still at UFES, Guidolini was involved in the development of a particularly useful and relevant technology: a tracker that identifies pedestrians at intersections and momentarily follows their movement to predict their future position, as detailed in a November 2019 article in Computers & Graphics<\/em>.<\/p>\n The following year, before founding Lume Robotics, he was part of a joint UFES and Embraer team that carried out the first test of an autonomous aircraft in Brazil: the Legacy 500 jet. \u201cThe objective was to prove that the system was capable of directing the plane along an airport runway,\u201d he says. \u201cWe installed sensors that capture information from the environment and computers to process this data, and the plane taxied without human intervention.\u201d<\/p>\n The autonomous vehicles operate using two systems\u2014one for perception and the other for decision-making\u2014developed at the startup and at the University of Bras\u00edlia (UnB), the University of S\u00e3o Paulo (USP), the Federal University of Esp\u00edrito Santo, the Federal University of Minas Gerais (UFMG), and the Federal University of Rio Grande do Sul (UFRGS) (see <\/em>Pesquisa FAPESP issue n\u00ba 315<\/em><\/a>).<\/p>\n The perception system consists of cameras, sensors, radars, and satellite navigation instruments, installed in different parts of the vehicle, as described in a March 2021 review article in the journal Expert Systems with Applications<\/em>, written by Guidolini and colleagues. Together, these devices perform essential tasks such as vehicle localization, road mapping, obstacle detection, and traffic-sign recognition.<\/p>\n The decision-making platform, equipped with artificial intelligence software, interprets the data gathered by the perception system, plans the route accordingly, and prevents accidents, allowing the vehicle to reach the destination defined by the user. The types and quantity of devices vary depending on the vehicle and the tasks it must perform.<\/p>\n Automation scale Trucks and buses operating experimentally fall between levels 3 and 4\u2014in the latter, the vehicle travels on restricted routes at speeds of up to 50 kilometers per hour (km\/h). In late 2024, the Vit\u00f3ria-based startup obtained R$2.5 million in funding from the Brazilian Funding Authority for Studies and Projects (FINEP) to create a more advanced version, not only driverless, but also cabinless. The first prototype is expected to be ready in 2027, and if everything goes well, field tests will begin in 2028.<\/p>\n Lume Robotics<\/span>Driverless bus was tested at a steel mill in Vit\u00f3ria, Esp\u00edrito Santo; below, a close-up of the Lidar sensorLume Robotics<\/span><\/p><\/div>\n \u201cLume is one of the most advanced companies in mobile robotics in Brazil. The first company to reach automation level 4 will put themselves in a good position to dominate the market,\u201d says Fernando Santos Os\u00f3rio, a computer scientist from the Mobile Robotics Laboratory at USP\u2019s Institute of Mathematical and Computer Sciences (LRM-ICMC), S\u00e3o Carlos. In some cities in the USA and Japan, says the researcher, there are already driverless cars traveling on routes planned and mapped by satellites. \u201cNone of them have yet achieved the maximum level of automation: level 5,\u201d says Os\u00f3rio.<\/p>\n There are still obstacles to overcome before vehicles of this type hit the streets of Brazil, including a need for greater technological maturity and various certification and legal issues. It is yet to be defined, for example, who should be held responsible if a fully automated vehicle runs a red light and hits a person or destroys the wall of a house.<\/p>\n Despite this fact, the autonomous vehicle market is growing. An August 2024 analysis by American consultancy Goldman Sachs estimated that driverless vehicles with intermediate levels of automation will account for around 60% of all light vehicle sales in 2040.<\/p>\n Sales are expected to be highest in China, where level 4 automated cars, such as Baidu subsidiary Apollo Go\u2019s robotaxis, are already circulating in some cities, including Shenzhen and Wuhan. General Motors, meanwhile, has abandoned development of its app-based robotaxi in the face of financial problems. In January, the American multinational announced Super Cruise, a driver assistance technology similar to Tesla\u2019s Autopilot. Both systems are designed to partially automate driving, using cameras, radar, and an attention module to ensure the driver is focused on the road.<\/p>\n \u201cIn other countries, autonomous cars are used in more structured cities, where the roads are in good condition and traffic flow is more organized than in Brazil,\u201d warns cartographic engineer Edvaldo Sim\u00f5es da Fonseca J\u00fanior, head of the Topography and Geodesy Laboratory at USP\u2019s Polytechnic School, who has been studying the topic since the early 2010s. To illustrate the challenges of implementing this technology in Brazilian cities, he recalls a situation he presented to students during a class: \u201cThe other day I took a photo of a traffic light that showed both a green and a red light at the same time. How does an autonomous car resolve a situation like this?\u201d The vehicle’s decision will depend on its programming, says Fonseca. \u201cIf it prioritizes looking for a red light, it will be stuck at the intersection forever. If its looks for green, it will drive forward and a collision could occur. If the car identifies that both lights are on, it will approach the intersection and use other sensors to check whether it should proceed or not.\u201d<\/p>\n Project carina Terra is head of Brazil\u2019s National Institute of Science and Technology for Cooperative Autonomous Systems (InSAC), one of the National Centers for Science, Technology, and Innovation (INCT) funded by FAPESP and the Brazilian National Council for Scientific and Technological Development (CNPq). InSAC succeeded the INCT for Critical Embedded Systems (INCT-SEC), which closed in 2014. Both of these INCTs provided essential support for the development of the Intelligent Robotic Car for Autonomous Navigation (Carina), the first automobile authorized to travel without a driver in a Brazilian city (S\u00e3o Carlos, in S\u00e3o Paulo State) in 2013. \u201cWe bought the first vehicle in 2009, with funding from INCT-SEC, and it was automated and operational, without a driver, in 2010,\u201d recalls Os\u00f3rio.<\/p>\n In Brazil, an additional obstacle for autonomous vehicle projects is the high cost of imported components. Os\u00f3rio hopes to help solve this problem through his participation in the project \u201cImplementation and testing of driving assistance components and devices\u201d (SEGCOM), which is part of the federal government’s Green Mobility and Innovation (MOVER) program. \u201cThe aim of the SEGCOM project is to identify technological alternatives that can be developed domestically,\u201d he explains. One of the goals is to create a computer vision system that uses cameras made by Brazilian company Intelbras.<\/p>\n \u201cWe are gradually incorporating technologies,\u201d says Terra. According to Os\u00f3rio, it is still difficult to predict when robotic cars will make it onto the streets of Brazilian cities\u2014not only for technical reasons, but also due to the lack of legislation authorizing this type of vehicle to operate. In search of alternatives, teams from USP\u2019s EESC and ICMC are working in partnership with Scania to collect the data needed to train an autonomous truck for mining activities. \u201cRealistically,\u201d Terra concludes, \u201cthis is the niche of restricted operations that is viable for now.\u201d<\/p>\n The story above was published with the title “No driver required<\/strong>” in issue in issue 349 of march\/2025.<\/p>\n Project<\/strong> Scientific articles
\n<\/strong>Vehicle automation levels range from zero, when the car, bus, or truck is completely dependent on the driver, to five, when the vehicle is fully automated, requiring no human intervention, regardless of the place or situation in which they operate. The classification system was developed by the US Society of Automotive Engineers (SAE).<\/p>\n![\"\"](\"https:\/\/revistapesquisa.fapesp.br\/wp-content\/uploads\/2025\/09\/rpf-lume-robotics-detalhe-lidar-349-1140b.jpg\")
\n<\/strong>For Marco Henrique Terra, an electrical engineer from USP\u2019s S\u00e3o Carlos School of Engineering (EESC), the biggest barrier to autonomous vehicle research in Brazil is not the lack of infrastructure. \u201cThe major bottleneck is capital. We need investors with the staying power to support the development of this technology until it reaches maturity.\u201d<\/p>\n
\nINCT 2014:<\/em> Brazilian National Science and Technology Institute for Autonomous Cooperative Systems Applied to Security and the Environment<\/em> (n\u00ba 14\/50851-0<\/u><\/a>); Grant Mechanism <\/strong>Thematic Project; Principal Investigator <\/strong>Marco Henrique Terra (USP); Investment<\/strong> R$4,227,952.62.<\/p>\n
\n<\/strong>BADUE, C. et al<\/em>. Self-driving cars: A survey<\/a>. Expert Systems with Applications<\/strong>. Vol. 165, 113816. Mar. 1, 2021.
\nSARCINELLI, R. et al<\/em>. Handling pedestrians in self-driving cars using image tracking and alternative path generation with Fren\u00e9t frames<\/a>. Computers & Graphics<\/strong>. Vol. 84, pp. 173\u201384. Nov. 2019.<\/p>\n","protected":false},"excerpt":{"rendered":"Autonomous navigation system developed by Vit\u00f3ria-based startup allows vehicles to identify obstacles and travel on predefined routes","protected":false},"author":131,"featured_media":563367,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_exactmetrics_skip_tracking":false,"_exactmetrics_sitenote_active":false,"_exactmetrics_sitenote_note":"","_exactmetrics_sitenote_category":0,"footnotes":""},"categories":[169],"tags":[243,2413],"coauthors":[440],"class_list":["post-563366","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-technology","tag-innovation","tag-technology"],"acf":[],"_links":{"self":[{"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/posts\/563366","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/users\/131"}],"replies":[{"embeddable":true,"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/comments?post=563366"}],"version-history":[{"count":2,"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/posts\/563366\/revisions"}],"predecessor-version":[{"id":563384,"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/posts\/563366\/revisions\/563384"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/media\/563367"}],"wp:attachment":[{"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/media?parent=563366"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/categories?post=563366"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/tags?post=563366"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/revistapesquisa.fapesp.br\/en\/wp-json\/wp\/v2\/coauthors?post=563366"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}