TNO Appl.AI is the branding for AI activities in the TNO units. Among these activities are TNO-wide science and technology research investments to further the capabilities and propositions that TNO can offer to its customers, research partners and other stakeholders. The goal of this Research Strategy document is to guide the investment decisions with a planning horizon of up to 10 – 15 years. Such decisions obviously also depend on the pace of developments outside TNO. This research strategy and an accompanying AI landscaping study will be used as input for proposals for AI research programs with specific timelines and performance goals of capabilities. These programs will be developed by TNO, whenever possible in collaboration with research partners. AI offers an enormous economic potential in terms of improving the effectiveness and efficiency of products and services for governments and companies. In applications that are critical with respect to safety, security or ethics, the use of AI is highly demanding. Acceptance by society will be crucial for the market prospects of AI-enabled systems. To TNO the European values with respect to trustworthy AI are therefore guiding principles. At TNO, AI technology developments focussed on adaptive AI are driven along four dimensions that define the complexity of deployment in the real-world: open environment, multiple purpose, high risk and teaming. The current document describes three roadmaps along which TNO intends to develop its Trustworthy adaptive AI capabilities. The development along these roadmaps is traceable by intermediate milestones. ROADMAP AUTONOMOUS SYSTEMS Autonomous systems gradually move from navigation and information collection in the environment to physical interventions and collaboration with other autonomous systems. These autonomous systems need to know their own and each other’s capabilities and limitations to balance control between the users and the systems. ROADMAP FEDERATED DECISION MAKING Humans making complex decisions can benefit from supportive systems. These systems should continuously and concurrently advise and learn from various types of users. Federation enables the users to remain in control and responsible. ROADMAP AI SYSTEMS ENGINEERING & LIFECYLE MANAGEMENT AI systems learn continuously and are thus fundamentally different to build and maintain than other (more static) software systems. This roadmap aims at developing capabilities for building, controlling, and managing AI systems with the functional capabilities developed in the other two roadmaps. The three roadmaps focus on capability development to support both short- and long-term programs, projects, and propositions. The outside world will have its own dynamics in challenges, limitations, and possibilities. TNO is well rooted in markets, society, government, and academics, that is, well positioned to monitor, co-design, co-develop, and adapt for future Trustworthy AI systems in high-risk application domains. In close cooperation with others TNO integrates different technologies in real life demonstrators and prototypes. Our aim is to deliver generic methods, techniques, and tools. Through publications, open source, and re-use for our stakeholders we disseminate and valorise this knowledge base.