Home

Tools for Finding Research Papers

In every research field, there are countless papers published each year. In my area, robotics, I categorize papers into four groups based on their research direction:

• My Current Research Area - Topics directly related to my ongoing work in robotics.
• Related Research Areas - Fields that are closely connected, such as AI/ML and Vision.
• Most Popular Research Areas - Areas that reflect industry needs or recent breakthroughs.
• Basic Research Areas - Foundational disciplines like mathematics and physics.

This categorization helps me focus on what's relevant while staying informed about broader developments.

Current Research Area - Robotics: My research focuses on robotics. To stay updated, I regularly review all papers published in key journals and conferences such as RSS, TRO, IJRR, and Science Robotics. For conferences like ICRA and IROS, which feature a large number of papers, I narrow my search by using keywords such as motion planning, swarm, multi-agent, field, or unstructured environments.

For papers that seem essential, I use tools like Connected Papers to explore related works, trace the origins of an idea, and check for follow-up research.

Related Research Areas - AI/ML and Vision: For related fields like AI/ML and Vision, I focus on understanding trends and key developments. These fields often prioritize conferences over journals, so I pay close attention to major conferences, including AAAI, ICLR, ICML, NeurIPS, 3DV, CVPR, ECCV, ICCV, and WACV.

Given the sheer volume of papers, it's impossible to read all of them—or even skim their titles. Instead, I take the following approach:

• I start by reviewing the Best Paper and Best Paper Candidate for each conference. These papers often represent cutting-edge research and highlight the conference's focus.
• I use tools like Paper Copilot to find highly-rated accepted papers. I skim the titles of around first 150 papers and read those that are directly relevant or interesting to me. There are also some useful visualization tools, such as visualization of the development of research directions.

Of course, there are still a lot of papers mey have someting we may interested. For these parts, I seek the help from social media and academic communities. For example, I follow IEEE Robotics and Automation Society groups and prominent researchers on platforms like X (formerly Twitter) and LinkedIn.

Most Popular and Basic Research Area: For popular research areas and foundational disciplines, my approach is less structured. I often browse sources like Nature News and Science Daily. Tools like OpenAlex are also useful for exploring new methods or topics.

I don't depend heavily on recommendation algorithms, as I prefer maintaining control over what I read. For organizing papers, I use the file system of the operating system. It's easy to back up.

It's easy to feel overwhelmed by the sheer number of papers and the fast pace of academic and industrial research. I remind myself that it's okay to miss some papers—important ones will likely resurface in the future.

Good luck with your research!