AI is transforming the physical world, and Xidi Autonomous Driving is advancing toward 'heavy-load embodied intelligence'

While the global AI wave focuses on algorithmic iterations in the digital realm, a quiet transformation is underway—one that redefines productivity in the physical world.

At the 2026 Asia Logistics Biennale, Sidewalk Intelligence $CIDI (03881.HK)$ held a press event, systematically showcasing its core achievements toward 'heavy-duty embodied intelligence,' unveiling TAG—the industry’s first portable V2P (Vehicle-to-Pedestrian) smart terminal—and signing cooperation agreements with Proton Auto and DeepRobotics.

Send machines to the most dangerous places

Every leap in human civilization’s productivity has begun with the emergence of a new tool: the steam engine enabled industrialized production, the internal combustion engine powered globalization, and computers ushered in the information age. From liberating physical labor to upgrading motive power and then to the computing revolution, each leap fundamentally extended the boundaries of human capability.

Standing at this new inflection point, what will be the next tool to transform the physical world?

Sidewalk Intelligence’s answer: heavy-duty embodied intelligence.

Heavy-duty embodied intelligence refers to equipping heavy machinery with autonomous operational capabilities through intelligent technologies.Heavy machinery is characterized by high tonnage, high payload capacity, and adaptability to extreme working conditions. It is deployed in scenarios such as rock breaking and drilling, explosive loading, excavation and loading, and heavy-load transportation, making it the most robust production tool. Physical AI endows heavy machinery with a complete perception system, decision-making 'brain,' and execution capabilities, enabling it to operate independently in extremely complex production environments and collaborate with other machines to accomplish integrated, multi-task scenarios.

Understand it with a simple formula:Heavy-duty embodied intelligence = Physical AI × Heavy machinery × High-risk operations.

Physical AI serves as the technological engine, heavy machinery as the physical carrier, and high-risk operations as the application scenario. These three components act as multiplicative factors, amplifying overall value.

Behind this strategic direction lies CIDI Smart Driving’s deep-rooted robotics heritage. Under the leadership of Professor Li Zexiang, the team has built a systematic robotics knowledge framework grounded in rigorous mathematical and physical logic. Their related monographs have become essential theoretical references in the global robotics community and serve as a foundational pillar and innovative core for the company’s technology research and product development.

Leveraging this technological foundation, CIDI Smart Driving is advancing its heavy-duty embodied intelligence strategy along three fronts:Heavy-duty autonomous driving(Transport robots)，Overcoming safety and efficiency bottlenecks in extreme operating conditions through autonomous mining trucks and autonomous heavy-duty trucks;Heavy Equipment Embodied Intelligence(Operational Robots)，Multiple types of embodied intelligent agents covering full-scenario operations from surface to underground environments;Intelligent Mobility,Built on vehicle-grade high-performance perception technology, addressing diverse safety and intelligent upgrade requirements across multiple scenarios.

"Autonomous driving is essentially a self-navigating wheeled robot. CIDI Smart Driving has already accumulated core robotic technologies and data in heavy-load control, extreme-condition perception, and fleet scheduling. Extending from wheeled robots to excavation, drilling, and explosive-loading robots is a natural progression along the same technological trajectory."Dr. Hu Sibo, CEO of CIDI Smart Driving,stated at the launch event.

▲ Dr. Hu Sibo, CEO of CIDI Smart Driving, delivers a keynote speech

This innovative technology is continuously being validated in real-world commercial applications. Currently, CIDI Smart Driving’s range of intelligent products has been deployed at scale inResource extraction and transportation, logistics, large-scale infrastructure, and transportation engineeringsectors, with total shipments exceeding20,000 units. By replacing or enhancing hazardous and labor-intensive manual operations, it continues to unlock deeper operational efficiencies in mining and industrial operations, driving a generational leap in heavy-load productivity.

In fiscal year 2025, Xidi Zhijia reported a 115.8% year-over-year increase in total revenue, reaching RMB 885 million, and an 86.7% year-over-year increase in gross profit, rapidly approaching its profitability inflection point.

▲ Xidi Zhijia's autonomous container trucks passing through the Puzhai–Xin Qing freight-only checkpoint on the China-Vietnam border

Becoming the 'brain' of heavy-load embodied intelligence

In terms of business model, Xidi Zhijia positions itself as the 'brain' of heavy-load embodied intelligence, focusing on technology enablement and serving as the industry’s most trusted technology partner.

Based on this positioning, Xidi Zhijia has built a"Heavy-Load Embodied Intelligence Technology Platform", structured in three layers.

The bottom layer is a unified data engine.It builds a standardized data foundation using production, environmental, and vehicle data from diverse operational scenarios.

The multimodal perception fusion system maintains continuous detection capability for both static and dynamic obstacles across the entire operational area—even under harsh conditions such as dust, vibration, extreme weather, and varying lighting. Its multi-source, heterogeneous environmental data continuously enhances the scenario coverage and model training sample quality of the data foundation.

The middle layer is the heavy-duty world model.Leveraging an edge-cloud collaborative architecture, the on-device action model enables real-time decision-making at millisecond latency, while the cloud-based training model continuously iterates and evolves.

Distinct from other types of world models, the heavy-duty world model deeply integrates domain-specific operational data, heavy-duty robot dynamics, and action-space reasoning, and is natively designed for multi-agent collaboration. Its core strength lies in understanding the bidirectional interaction between agents and their physical environment—continuously inferring the cumulative impact of an agent’s own actions on the environment and generating precise plans based on the resulting evolved environmental state. This closed-loop decision-making capability, centered on environmental evolution reasoning, forms a unique technological barrier specific to heavy-duty operational scenarios.

The top layer is the cluster decision-making and scheduling system.Aiming for global optimality, it coordinates heterogeneous robots to operate collaboratively within the same spatiotemporal domain. This highly mature multi-type swarm intelligence solution has already undergone successful engineering validation in complex mining environments characterized by dense, mixed-agent operations.

The key advantage of this integrated architecture lies in its exceptional generalization capability: the same system can be adapted into multiple configurations and reused across different scenarios. Model capabilities developed in one scenario can be transferred to others at low cost, enabling continuous realization of technological value.

At the press conference, CIDI Autonomous Driving also signed cooperation agreements with Proton Motors and DeepRobotics respectively.

Proton Motors is a vehicle manufacturer under Shaanxi Automobile Group, specializing in new energy intelligent connected commercial vehicles, and is continuously innovating in four key technological areas: intelligentization, electrification, connectivity, and lightweighting of commercial vehicles.CIDI Autonomous Driving and Proton Motors have reached a deep strategic collaboration, focusing on comprehensive solutions for autonomous heavy-duty trucks to jointly advance the adoption and application of autonomous driving technology in commercial vehicle scenarios.

DeepRobotics specializes in the research, development, and industrialization of quadruped and humanoid robots, and has already achieved commercial deployment in sectors such as power, firefighting, and security.The partnership between the two parties will focus on mining robots.

The collaborations with Proton Motors and DeepRobotics represent a concentrated demonstration of CIDI Autonomous Driving’s ecosystem synergy strategy. From signing agreements with Sinoma Mining and Sinoma Institute of Mining in September 2025, to securing a purchase order for 500 autonomous mining trucks from Guangna Group in February 2026, followed by partnerships with the UK-based MMD Group and Schaeffler in April, and then consecutive collaborations in June with Yipuli, Zoomlion, General Consulting, Proton Motors, and DeepRobotics, CIDI Autonomous Driving’s 'circle of partners' is rapidly expanding along"vehicle platforms—application scenarios—global expansion"three strategic tracks.

Its open positioning as a developer of 'brains' for heavy-duty embodied intelligence is attracting an increasing number of partners across the industry chain.

Expanding from China to global markets

Heavy-duty embodied intelligence represents a trillion-dollar market driven by both policy and market forces. On the policy front, the principle of 'safety through reduced human presence' has become a clear strategic direction, with safety imperatives now serving as a top-down catalyst for industrial upgrading. On the market side, AI demands computing power, and computing power requires energy. Heavy-duty embodied intelligence sits precisely at the intersection of the AI revolution and energy infrastructure, combining the explosive growth potential of technology with the certainty of the energy sector.

The window of opportunity has already opened. Taking mining—the largest commercial deployment scenario—as an example, the current penetration rate of autonomous mining trucks stands at around 10%, and the inflection point for large-scale adoption is drawing near. China’s mature new energy vehicle supply chain is comprehensively empowering embodied intelligence, placing Xidi Zhijia squarely on the crest of this industrial acceleration wave.

An even broader vision lies in global expansion. As one of the first Chinese companies to achieve commercial validation, Xidi Zhijia is bringing China’s heavy-duty embodied intelligence advantages to the global stage.

Xidi Zhijia has already initiated its global rollout targeting resource-rich countries such as Australia, Brazil, and Indonesia, establishing partnerships with multiple mine operators, OEMs, and service providers. The company plans to progressively expand into markets across Africa, Central Asia, and the Middle East over the next few years, achieving commercial-scale deliveries and securing a leading position in overseas heavy-duty applications.

As a form of new quality productive forces, heavy-duty embodied intelligence is transforming the most fundamental modes of physical-world production. Xidi Zhijia focuses on operational scenarios characterized by high risk, labor intensity, and high value—domains protected by deep professional barriers. Through systematic innovation, Xidi Zhijia aims to liberate humans from hazardous and strenuous work environments, deploying sustainable intelligent transport capabilities to reshape the physical world and drive a paradigm shift in industrial civilization.

$NVIDIA (NVDA.US)$ $BLACK SESAME (02533.HK)$ $ONEROBOTICS (06600.HK)$

9078 Views